/*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
*/
/*
* CoverTree.java
* Copyright (C) 2006 Alina Beygelzimer and Sham Kakade and John Langford
*/
package weka.core.neighboursearch;
import weka.core.DistanceFunction;
import weka.core.EuclideanDistance;
import weka.core.Instance;
import weka.core.Instances;
import weka.core.Option;
import weka.core.RevisionHandler;
import weka.core.RevisionUtils;
import weka.core.TechnicalInformation;
import weka.core.TechnicalInformationHandler;
import weka.core.Utils;
import weka.core.TechnicalInformation.Field;
import weka.core.TechnicalInformation.Type;
import weka.core.converters.CSVLoader;
import weka.core.neighboursearch.covertrees.Stack;
import java.io.BufferedReader;
import java.io.File;
import java.io.FileReader;
import java.io.Serializable;
import java.util.Enumeration;
import java.util.List;
import java.util.Vector;
/**
<!-- globalinfo-start -->
* Class implementing the CoverTree datastructure.<br/>
* The class is very much a translation of the c source code made available by the authors.<br/>
* <br/>
* For more information and original source code see:<br/>
* <br/>
* Alina Beygelzimer, Sham Kakade, John Langford: Cover trees for nearest neighbor. In: ICML'06: Proceedings of the 23rd international conference on Machine learning, New York, NY, USA, 97-104, 2006.
* <p/>
<!-- globalinfo-end -->
*
<!-- technical-bibtex-start -->
* BibTeX:
* <pre>
* @inproceedings{Beygelzimer2006,
* address = {New York, NY, USA},
* author = {Alina Beygelzimer and Sham Kakade and John Langford},
* booktitle = {ICML'06: Proceedings of the 23rd international conference on Machine learning},
* pages = {97-104},
* publisher = {ACM Press},
* title = {Cover trees for nearest neighbor},
* year = {2006},
* location = {Pittsburgh, Pennsylvania},
* HTTP = {http://hunch.net/\~jl/projects/cover_tree/cover_tree.html}
* }
* </pre>
* <p/>
<!-- technical-bibtex-end -->
*
<!-- options-start -->
* Valid options are: <p/>
*
* <pre> -B <value>
* Set base of the expansion constant
* (default = 1.3).</pre>
*
<!-- options-end -->
*
* @author Alina Beygelzimer (original C++ code)
* @author Sham Kakade (original C++ code)
* @author John Langford (original C++ code)
* @author Ashraf M. Kibriya (amk14[at-the-rate]cs[dot]waikato[dot]ac[dot]nz) (Java port)
* @version $Revision: 1.4 $
*/
public class CoverTree
extends NearestNeighbourSearch
implements TechnicalInformationHandler {
/** for serialization. */
private static final long serialVersionUID = 7617412821497807586L;
/**
* class representing a node of the cover tree.
*
* @author Ashraf M. Kibriya (amk14[at-the-rate]cs[dot]waikato[dot]ac[dot]nz)
* @version $Revision: 1.4 $
*/
public class CoverTreeNode
implements Serializable, RevisionHandler {
/** for serialization. */
private static final long serialVersionUID = 1808760031169036512L;
/** ID for the node. */
private int nodeid;
/** Index of the instance represented by this node in the index array. */
private Integer idx;
/** The distance of the furthest descendant of the node. */
private double max_dist; // The maximum distance to any grandchild.
/** The distance to the nodes parent. */
private double parent_dist; // The distance to the parent.
/** The children of the node. */
private Stack<CoverTreeNode> children;
/** The number of children node has. */
private int num_children; // The number of children.
/** The min i that makes base^i <= max_dist. */
private int scale; // Essentially, an upper bound on the distance to any child.
/** Constructor for the class. */
public CoverTreeNode() {
}
/**
* Constructor.
* @param i The index of the Instance this node is
* associated with.
* @param md The distance of the furthest descendant.
* @param pd The distance of the node to its parent.
* @param childs Children of the node in a stack.
* @param numchilds The number of children of the
* node.
* @param s The scale/level of the node in the tree.
*/
public CoverTreeNode(Integer i, double md, double pd,
Stack<CoverTreeNode> childs, int numchilds, int s) {
idx = i;
max_dist = md;
parent_dist = pd;
children = childs;
num_children = numchilds;
scale = s;
}
/** Returns the instance represented by the node.
* @return The instance represented by the node.
*/
public Instance p() {
return m_Instances.instance(idx);
}
/** Returns whether if the node is a leaf or not.
* @return true if the node is a leaf node.
*/
public boolean isALeaf() {
return num_children==0;
}
/**
* Returns the revision string.
*
* @return the revision
*/
public String getRevision() {
return RevisionUtils.extract("$Revision: 1.4 $");
}
}
/**
* Private class holding a point's distance to the current reference
* point p.
*
* @author Ashraf M. Kibriya (amk14[at-the-rate]cs[dot]waikato[dot]ac[dot]nz)
* @version $Revision: 1.4 $
*/
private class DistanceNode
implements RevisionHandler {
/**
* The last distance is to the current reference point
* (potential current parent). The previous ones are
* to reference points that were previously looked at
* (all potential ancestors).
*/
Stack<Double> dist;
/** The index of the instance represented by this node. */
Integer idx;
/**
* Returns the instance represent by this DistanceNode.
* @return The instance represented by this node.
*/
public Instance q() {
return m_Instances.instance(idx);
}
/**
* Returns the revision string.
*
* @return the revision
*/
public String getRevision() {
return RevisionUtils.extract("$Revision: 1.4 $");
}
}
/** The euclidean distance function to use. */
protected EuclideanDistance m_EuclideanDistance;
{ // to make sure we have only one object of EuclideanDistance
if (m_DistanceFunction instanceof EuclideanDistance)
m_EuclideanDistance = (EuclideanDistance) m_DistanceFunction;
else
m_DistanceFunction = m_EuclideanDistance = new EuclideanDistance();
}
/** The root node. */
protected CoverTreeNode m_Root;
/**
* Array holding the distances of the nearest neighbours. It is filled up
* both by nearestNeighbour() and kNearestNeighbours().
*/
protected double [] m_DistanceList;
/** Number of nodes in the tree. */
protected int m_NumNodes, m_NumLeaves, m_MaxDepth;
/** Tree Stats variables. */
protected TreePerformanceStats m_TreeStats = null;
/**
* The base of our expansion constant. In other words the 2 in 2^i used
* in covering tree and separation invariants of a cover tree. P.S.: In
* paper it's suggested the separation invariant is relaxed in batch
* construction.
*/
protected double m_Base = 1.3;
/**
* if we have base 2 then this can be viewed as 1/ln(2), which can be
* used later on to do il2*ln(d) instead of ln(d)/ln(2), to get log2(d),
* in get_scale method.
*/
protected double il2 = 1.0 / Math.log(m_Base);
/**
* default constructor.
*/
public CoverTree() {
super();
if(getMeasurePerformance())
m_Stats = m_TreeStats = new TreePerformanceStats();
}
/**
* Returns a string describing this nearest neighbour search algorithm.
*
* @return a description of the algorithm for displaying in the
* explorer/experimenter gui
*/
public String globalInfo() {
return
"Class implementing the CoverTree datastructure.\n"
+ "The class is very much a translation of the c source code made "
+ "available by the authors.\n\n"
+ "For more information and original source code see:\n\n"
+ getTechnicalInformation().toString();
}
/**
* Returns an instance of a TechnicalInformation object, containing detailed
* information about the technical background of this class, e.g., paper
* reference or book this class is based on.
*
* @return the technical information about this class
*/
public TechnicalInformation getTechnicalInformation() {
TechnicalInformation result;
result = new TechnicalInformation(Type.INPROCEEDINGS);
result.setValue(Field.AUTHOR, "Alina Beygelzimer and Sham Kakade and John Langford");
result.setValue(Field.TITLE, "Cover trees for nearest neighbor");
result.setValue(Field.BOOKTITLE, "ICML'06: Proceedings of the 23rd international conference on Machine learning");
result.setValue(Field.PAGES, "97-104");
result.setValue(Field.YEAR, "2006");
result.setValue(Field.PUBLISHER, "ACM Press");
result.setValue(Field.ADDRESS, "New York, NY, USA");
result.setValue(Field.LOCATION, "Pittsburgh, Pennsylvania");
result.setValue(Field.HTTP, "http://hunch.net/~jl/projects/cover_tree/cover_tree.html");
return result;
}
/**
* Returns an enumeration describing the available options.
*
* @return an enumeration of all the available options.
*/
public Enumeration listOptions() {
Vector newVector = new Vector();
newVector.addElement(new Option(
"\tSet base of the expansion constant\n"
+ "\t(default = 1.3).",
"B", 1, "-B <value>"));
return newVector.elements();
}
/**
* Parses a given list of options. <p/>
*
<!-- options-start -->
* Valid options are: <p/>
*
* <pre> -B <value>
* Set base of the expansion constant
* (default = 1.3).</pre>
*
<!-- options-end -->
*
* @param options the list of options as an array of strings
* @throws Exception if an option is not supported
*/
public void setOptions(String[] options)
throws Exception {
super.setOptions(options);
String optionString = Utils.getOption('B', options);
if (optionString.length() != 0)
setBase(Double.parseDouble(optionString));
else
setBase(1.3);
}
/**
* Gets the current settings of KDtree.
*
* @return an array of strings suitable for passing to setOptions
*/
public String[] getOptions() {
Vector<String> result;
String[] options;
int i;
result = new Vector<String>();
options = super.getOptions();
for (i = 0; i < options.length; i++)
result.add(options[i]);
result.add("-B");
result.add("" + getBase());
return result.toArray(new String[result.size()]);
}
/**
* Returns the distance/value of a given scale/level. I.e. the value of
* base^i (e.g. 2^i).
*
* @param s the level/scale
* @return base^s
*/
protected double dist_of_scale(int s) {
return Math.pow(m_Base, s);
}
/**
* Finds the scale/level of a given value. I.e. the "i" in base^i.
*
* @param d the value whose scale/level is to be determined.
* @return the scale/level of the given value.
*/
protected int get_scale(double d) {
return (int) Math.ceil(il2 * Math.log(d));
}
/**
* Creates a new internal node for a given Instance/point p.
* @param idx The index of the instance the node represents.
* @return Newly created CoverTreeNode.
*/
protected CoverTreeNode new_node(Integer idx) { // const point &p)
CoverTreeNode new_node = new CoverTreeNode();
new_node.idx = idx;
return new_node;
}
/**
* Creates a new leaf node for a given Instance/point p.
* @param idx The index of the instance this leaf node
* represents.
* @return Newly created leaf CoverTreeNode.
*/
protected CoverTreeNode new_leaf(Integer idx) { // (const point &p)
CoverTreeNode new_leaf = new CoverTreeNode(idx, 0.0, 0.0, null, 0, 100);
return new_leaf;
}
/**
* Returns the max distance of the reference point p in current node to
* it's children nodes.
* @param v The stack of DistanceNode objects.
* @return Distance of the furthest child.
*/
protected double max_set(Stack<DistanceNode> v) { // rename to
// maxChildDist
double max = 0.0;
for (int i = 0; i < v.length; i++) {
DistanceNode n = v.element(i);
if (max < n.dist.element(n.dist.length - 1).floatValue()) { // v[i].dist.last())
max = n.dist.element(n.dist.length - 1).floatValue(); // v[i].dist.last();
}
}
return max;
}
/**
* Splits a given point_set into near and far based on the given
* scale/level. All points with distance > base^max_scale would be moved
* to far set. In other words, all those points that are not covered by the
* next child ball of a point p (ball made of the same point p but of
* smaller radius at the next lower level) are removed from the supplied
* current point_set and put into far_set.
*
* @param point_set The supplied set from which all far points
* would be removed.
* @param far_set The set in which all far points having distance
* > base^max_scale would be put into.
* @param max_scale The given scale based on which the distances
* of points are judged to be far or near.
*/
protected void split(Stack<DistanceNode> point_set,
Stack<DistanceNode> far_set, int max_scale) {
int new_index = 0;
double fmax = dist_of_scale(max_scale);
for (int i = 0; i < point_set.length; i++) {
DistanceNode n = point_set.element(i);
if (n.dist.element(n.dist.length - 1).doubleValue() <= fmax) {
point_set.set(new_index++, point_set.element(i));
} else
far_set.push(point_set.element(i)); // point_set[i]);
}
List l = new java.util.LinkedList();
for (int i = 0; i < new_index; i++)
l.add(point_set.element(i));
//removing all and adding only the near points
point_set.clear();
point_set.addAll(l); // point_set.index=new_index;
}
/**
* Moves all the points in point_set covered by (the ball of) new_point
* into new_point_set, based on the given scale/level.
*
* @param point_set The supplied set of instances from which
* all points covered by new_point will be removed.
* @param new_point_set The set in which all points covered by
* new_point will be put into.
* @param new_point The given new point.
* @param max_scale The scale based on which distances are
* judged (radius of cover ball is calculated).
*/
protected void dist_split(Stack<DistanceNode> point_set,
Stack<DistanceNode> new_point_set,
DistanceNode new_point, int max_scale) {
int new_index = 0;
double fmax = dist_of_scale(max_scale);
for (int i = 0; i < point_set.length; i++) {
double new_d = Math.sqrt(m_DistanceFunction.distance(new_point.q(),
point_set.element(i).q(), fmax*fmax));
if (new_d <= fmax) {
point_set.element(i).dist.push(new_d);
new_point_set.push(point_set.element(i));
} else
point_set.set(new_index++, point_set.element(i));
}
List l = new java.util.LinkedList();
for (int i = 0; i < new_index; i++)
l.add(point_set.element(i));
point_set.clear();
point_set.addAll(l);
}
/**
* Creates a cover tree recursively using batch insert method.
*
* @param p The index of the instance from which to create the
* first node. All other points will be inserted beneath this node
* for p.
* @param max_scale The current scale/level where the node is to be
* created (Also determines the radius of the cover balls created at
* this level).
* @param top_scale The max scale in the whole tree.
* @param point_set The set of unprocessed points from which child nodes
* need to be created.
* @param consumed_set The set of processed points from which child
* nodes have already been created. This would be used to find the
* radius of the cover ball of p.
* @return the node of cover tree created with p.
*/
protected CoverTreeNode batch_insert(Integer p, int max_scale, // current
// scale/level
int top_scale, // max scale/level for this dataset
Stack<DistanceNode> point_set, // set of points that are nearer to p
// [will also contain returned unused
// points]
Stack<DistanceNode> consumed_set) // to return the set of points that have
// been used to calc. max_dist to a
// descendent
// Stack<Stack<DistanceNode>> stack) //may not be needed
{
if (point_set.length == 0) {
CoverTreeNode leaf = new_leaf(p);
leaf.nodeid = m_NumNodes;
m_NumNodes++; // incrementing node count
m_NumLeaves++; // incrementing leaves count
return leaf;
} else {
double max_dist = max_set(point_set); // O(|point_set|) the max dist
// in point_set to point "p".
int next_scale = Math.min(max_scale - 1, get_scale(max_dist));
if (next_scale == Integer.MIN_VALUE) { // We have points with distance
// 0. if max_dist is 0.
Stack<CoverTreeNode> children = new Stack<CoverTreeNode>();
CoverTreeNode leaf = new_leaf(p);
leaf.nodeid = m_NumNodes;
children.push(leaf);
m_NumLeaves++;
m_NumNodes++; // incrementing node and leaf count
while (point_set.length > 0) {
DistanceNode tmpnode = point_set.pop();
leaf = new_leaf(tmpnode.idx);
leaf.nodeid = m_NumNodes;
children.push(leaf);
m_NumLeaves++;
m_NumNodes++; // incrementing node and leaf count
consumed_set.push(tmpnode);
}
CoverTreeNode n = new_node(p); // make a new node out of p and assign
// it the children.
n.nodeid = m_NumNodes;
m_NumNodes++; // incrementing node count
n.scale = 100; // A magic number meant to be larger than all scales.
n.max_dist = 0; // since all points have distance 0 to p
n.num_children = children.length;
n.children = children;
return n;
} else {
Stack<DistanceNode> far = new Stack<DistanceNode>();
split(point_set, far, max_scale); // O(|point_set|)
CoverTreeNode child = batch_insert(p, next_scale, top_scale, point_set,
consumed_set);
if (point_set.length == 0) { // not creating any node in this
// recursive call
// push(stack,point_set);
point_set.replaceAllBy(far); // point_set=far;
return child;
} else {
CoverTreeNode n = new_node(p);
n.nodeid = m_NumNodes;
m_NumNodes++; // incrementing node count
Stack<CoverTreeNode> children = new Stack<CoverTreeNode>();
children.push(child);
while (point_set.length != 0) { // O(|point_set| * num_children)
Stack<DistanceNode> new_point_set = new Stack<DistanceNode>();
Stack<DistanceNode> new_consumed_set = new Stack<DistanceNode>();
DistanceNode tmpnode = point_set.pop();
double new_dist = tmpnode.dist.last();
consumed_set.push(tmpnode);
// putting points closer to new_point into new_point_set (and
// removing them from point_set)
dist_split(point_set, new_point_set, tmpnode, max_scale); // O(|point_saet|)
// putting points closer to new_point into new_point_set (and
// removing them from far)
dist_split(far, new_point_set, tmpnode, max_scale); // O(|far|)
CoverTreeNode new_child = batch_insert(tmpnode.idx, next_scale,
top_scale, new_point_set, new_consumed_set);
new_child.parent_dist = new_dist;
children.push(new_child);
// putting the unused points from new_point_set back into
// point_set and far
double fmax = dist_of_scale(max_scale);
tmpnode = null;
for (int i = 0; i < new_point_set.length; i++) { // O(|new_point_set|)
tmpnode = new_point_set.element(i);
tmpnode.dist.pop();
if (tmpnode.dist.last() <= fmax)
point_set.push(tmpnode);
else
far.push(tmpnode);
}
// putting the points consumed while recursing for new_point
// into consumed_set
tmpnode = null;
for (int i = 0; i < new_consumed_set.length; i++) { // O(|new_point_set|)
tmpnode = new_consumed_set.element(i);
tmpnode.dist.pop();
consumed_set.push(tmpnode);
}
}// end while(point_size.size!=0)
point_set.replaceAllBy(far); // point_set=far;
n.scale = top_scale - max_scale;
n.max_dist = max_set(consumed_set);
n.num_children = children.length;
n.children = children;
return n;
}// end else if(pointset!=0)
}// end else if(next_scale != -214....
}// end else if(pointset!=0)
}
/**
* Builds the tree on the given set of instances.
* P.S.: For internal use only. Outside classes
* should call setInstances().
* @param insts The instances on which to build
* the cover tree.
* @throws Exception If the supplied set of
* Instances is empty, or if there are missing
* values.
*/
protected void buildCoverTree(Instances insts) throws Exception {
if (insts.numInstances() == 0)
throw new Exception(
"CoverTree: Empty set of instances. Cannot build tree.");
checkMissing(insts);
if (m_EuclideanDistance == null)
m_DistanceFunction = m_EuclideanDistance = new EuclideanDistance(insts);
else
m_EuclideanDistance.setInstances(insts);
Stack<DistanceNode> point_set = new Stack<DistanceNode>();
Stack<DistanceNode> consumed_set = new Stack<DistanceNode>();
Instance point_p = insts.instance(0); int p_idx = 0;
double max_dist=-1, dist=0.0; Instance max_q=point_p;
for (int i = 1; i < insts.numInstances(); i++) {
DistanceNode temp = new DistanceNode();
temp.dist = new Stack<Double>();
dist = Math.sqrt(m_DistanceFunction.distance(point_p, insts.instance(i), Double.POSITIVE_INFINITY));
if(dist > max_dist) {
max_dist = dist; max_q = insts.instance(i);
}
temp.dist.push(dist);
temp.idx = i;
point_set.push(temp);
}
max_dist = max_set(point_set);
m_Root = batch_insert(p_idx, get_scale(max_dist), get_scale(max_dist),
point_set, consumed_set);
}
/*********************************NNSearch related stuff********************/
/**
* A class for a heap to store the nearest k neighbours to an instance.
* The heap also takes care of cases where multiple neighbours are the same
* distance away.
* i.e. the minimum size of the heap is k.
*
* @author Ashraf M. Kibriya (amk14[at-the-rate]cs[dot]waikato[dot]ac[dot]nz)
* @version $Revision: 1.4 $
*/
protected class MyHeap
implements RevisionHandler {
/** the heap. */
MyHeapElement m_heap[] = null;
/**
* constructor.
* @param maxSize the maximum size of the heap
*/
public MyHeap(int maxSize) {
if((maxSize%2)==0)
maxSize++;
m_heap = new MyHeapElement[maxSize+1];
m_heap[0] = new MyHeapElement(-1);
}
/**
* returns the size of the heap.
* @return the size
*/
public int size() {
return m_heap[0].index;
}
/**
* peeks at the first element.
* @return the first element
*/
public MyHeapElement peek() {
return m_heap[1];
}
/**
* returns the first element and removes it from the heap.
* @return the first element
* @throws Exception if no elements in heap
*/
public MyHeapElement get() throws Exception {
if(m_heap[0].index==0)
throw new Exception("No elements present in the heap");
MyHeapElement r = m_heap[1];
m_heap[1] = m_heap[m_heap[0].index];
m_heap[0].index--;
downheap();
return r;
}
/**
* adds the distance value to the heap.
*
* @param d the distance value
* @throws Exception if the heap gets too large
*/
public void put(double d) throws Exception {
if((m_heap[0].index+1)>(m_heap.length-1))
throw new Exception("the number of elements cannot exceed the "+
"initially set maximum limit");
m_heap[0].index++;
m_heap[m_heap[0].index] = new MyHeapElement(d);
upheap();
}
/**
* Puts an element by substituting it in place of
* the top most element.
*
* @param d The distance value.
* @throws Exception If distance is smaller than that of the head
* element.
*/
public void putBySubstitute(double d) throws Exception {
MyHeapElement head = get();
put(d);
if(head.distance == m_heap[1].distance) {
putKthNearest(head.distance);
}
else if(head.distance > m_heap[1].distance) {
m_KthNearest = null;
m_KthNearestSize = 0;
initSize = 10;
}
else if(head.distance < m_heap[1].distance) {
throw new Exception("The substituted element is greater than the "+
"head element. put() should have been called "+
"in place of putBySubstitute()");
}
}
/** the kth nearest ones. */
MyHeapElement m_KthNearest[] = null;
/** The number of kth nearest elements. */
int m_KthNearestSize = 0;
/** the initial size of the heap. */
int initSize=10;
/**
* returns the number of k nearest.
*
* @return the number of k nearest
* @see #m_KthNearestSize
*/
public int noOfKthNearest() {
return m_KthNearestSize;
}
/**
* Stores kth nearest elements (if there are
* more than one).
* @param d the distance
*/
public void putKthNearest(double d) {
if(m_KthNearest==null) {
m_KthNearest = new MyHeapElement[initSize];
}
if(m_KthNearestSize>=m_KthNearest.length) {
initSize += initSize;
MyHeapElement temp[] = new MyHeapElement[initSize];
System.arraycopy(m_KthNearest, 0, temp, 0, m_KthNearest.length);
m_KthNearest = temp;
}
m_KthNearest[m_KthNearestSize++] = new MyHeapElement(d);
}
/**
* returns the kth nearest element or null if none there.
*
* @return the kth nearest element
*/
public MyHeapElement getKthNearest() {
if(m_KthNearestSize==0)
return null;
m_KthNearestSize--;
return m_KthNearest[m_KthNearestSize];
}
/**
* performs upheap operation for the heap
* to maintian its properties.
*/
protected void upheap() {
int i = m_heap[0].index;
MyHeapElement temp;
while( i > 1 && m_heap[i].distance>m_heap[i/2].distance) {
temp = m_heap[i];
m_heap[i] = m_heap[i/2];
i = i/2;
m_heap[i] = temp; //this is i/2 done here to avoid another division.
}
}
/**
* performs downheap operation for the heap
* to maintian its properties.
*/
protected void downheap() {
int i = 1;
MyHeapElement temp;
while( ( (2*i) <= m_heap[0].index &&
m_heap[i].distance < m_heap[2*i].distance )
||
( (2*i+1) <= m_heap[0].index &&
m_heap[i].distance < m_heap[2*i+1].distance) ) {
if((2*i+1)<=m_heap[0].index) {
if(m_heap[2*i].distance>m_heap[2*i+1].distance) {
temp = m_heap[i];
m_heap[i] = m_heap[2*i];
i = 2*i;
m_heap[i] = temp;
}
else {
temp = m_heap[i];
m_heap[i] = m_heap[2*i+1];
i = 2*i+1;
m_heap[i] = temp;
}
}
else {
temp = m_heap[i];
m_heap[i] = m_heap[2*i];
i = 2*i;
m_heap[i] = temp;
}
}
}
/**
* returns the total size.
*
* @return the total size
*/
public int totalSize() {
return size()+noOfKthNearest();
}
/**
* Returns the revision string.
*
* @return the revision
*/
public String getRevision() {
return RevisionUtils.extract("$Revision: 1.4 $");
}
}
/**
* A class for storing data about a neighboring instance.
*
* @author Ashraf M. Kibriya (amk14[at-the-rate]cs[dot]waikato[dot]ac[dot]nz)
* @version $Revision: 1.4 $
*/
protected class MyHeapElement
implements RevisionHandler {
/** the distance. */
public double distance;
/**
* The index of this element. Also used as
* the size of the heap in the first element.
*/
int index = 0;
/**
* constructor.
*
* @param d the distance
*/
public MyHeapElement(double d) {
distance = d;
}
/**
* Returns the revision string.
*
* @return the revision
*/
public String getRevision() {
return RevisionUtils.extract("$Revision: 1.4 $");
}
}
/**
* stores a CoverTreeNode and its distance to the current query node.
*
* @author Ashraf M. Kibriya (amk14[at-the-rate]cs[dot]waikato[dot]ac[dot]nz)
* @version $Revision: 1.4 $
*/
private class d_node
implements RevisionHandler {
/** The distance of the node's point to the query point. */
double dist;
/** The node. */
CoverTreeNode n;
/**
* Constructor.
* @param d The distance of the node to the query.
* @param node The node.
*/
public d_node(double d, CoverTreeNode node) {
dist = d;
n = node;
}
/**
* Returns the revision string.
*
* @return the revision
*/
public String getRevision() {
return RevisionUtils.extract("$Revision: 1.4 $");
}
};
/**
* Initializes a heap with k values of the the given upper_bound.
*
* @param heap The heap to put values into.
* @param upper_bound The value to put into heap (the value with
* which it should be initialized).
* @param k The number of times upper_bound should be put into
* heap for initialization.
* @throws Exception If there is some problem in initializing
* the heap (if k > size of the heap).
*/
protected void setter(MyHeap heap, double upper_bound, final int k) throws Exception {
if(heap.size()>0)
heap.m_heap[0].index=0;
while(heap.size() < k) {
heap.put(upper_bound);
}
}
/**
* Replaces the current top/max value in the heap with the new one.
* The new max value should be <= the old one.
*
* @param upper_bound The heap.
* @param new_bound The new value that should replace the old top one.
* @throws Exception if the new value is greater than the old value.
*/
protected void update(MyHeap upper_bound, double new_bound) throws Exception {
upper_bound.putBySubstitute(new_bound);
}
/**
* Returns a cover set for a given level/scale.
* A cover set for a level consists of nodes whose
* Instances/centres are which are inside the query
* ball at that level. If no cover set exists for the
* given level (if it is the first time it is going
* to be used), than a new one is created.
*
* @param idx The level/scale for which the cover set
* is required.
* @param cover_sets The covers sets. Consists of stack
* of a stack of d_node objects.
* @return The cover set for the given level/scale.
*/
protected Stack<d_node> getCoverSet(int idx, Stack<Stack<d_node>> cover_sets) {
if (cover_sets.length <= idx) {
int i = cover_sets.length - 1;
while (i < idx) {
i++;
Stack<d_node> new_cover_set = new Stack<d_node>();
cover_sets.push(new_cover_set);
}
}
return cover_sets.element(idx);
}
/**
* Copies the contents of one zero set to the other. This
* is required if we are going to inspect child of some query node
* (if the queries are given in batch in the form of a cover tree).
* Only those nodes are copied to the new zero set that are inside
* the query ball of query_chi.
* P.S.: A zero set is a set of all leaf nodes that are found
* to be inside the query ball.
*
* @param query_chi The child node of our query node that we are
* going to inspect.
* @param new_upper_k New heap that will store the distances of the
* k NNs for query_chi.
* @param zero_set The zero set of query_chi's parent that needs
* to be copied.
* @param new_zero_set The new zero set of query_chi where old zero
* sets need to be copied into.
* @throws Exception If there is some problem.
*/
protected void copy_zero_set(CoverTreeNode query_chi, MyHeap new_upper_k,
Stack<d_node> zero_set, Stack<d_node> new_zero_set) throws Exception {
new_zero_set.clear();
d_node ele;
for (int i = 0; i < zero_set.length; i++) {
ele = zero_set.element(i);
double upper_dist = new_upper_k.peek().distance + query_chi.max_dist;
if (shell(ele.dist, query_chi.parent_dist, upper_dist)) {
double d = Math.sqrt(m_DistanceFunction.distance(query_chi.p(), ele.n
.p(), upper_dist * upper_dist));
if (m_TreeStats != null)
m_TreeStats.incrPointCount();
if (d <= upper_dist) {
if (d < new_upper_k.peek().distance)
update(new_upper_k, d);
d_node temp = new d_node(d, ele.n);
new_zero_set.push(temp);
if (m_TreeStats != null)
m_TreeStats.incrLeafCount();
}//end if(d<newupperbound)
}//end if(shell(...
}//end for
}
/**
* Copies the contents of one set of cover sets to the other. It
* is required if we are going to inspect child of some query node
* (if the queries are given in batch in the form of a cover tree).
* For each level, only those nodes are copied to the new set
* which are inside the query ball of query_chi at that level.
*
* @param query_chi The child node of our query node that we are
* going to inspect.
* @param new_upper_k New heap that will store the distances of the
* k NNs for query_chi.
* @param cover_sets The cover_sets of query_chi's parent, which
* need to be copied to new_cover_sets.
* @param new_cover_sets The new set of cover_sets that need to
* contain contents of cover_sets.
* @param current_scale The scale/level we are inspecting in our
* cover tree.
* @param max_scale The maximum level so far possible in our
* search (this is only updated as we descend and a deeper
* child is found inside the query ball).
* @throws Exception If there is problem.
*/
protected void copy_cover_sets(CoverTreeNode query_chi, MyHeap new_upper_k,
Stack<Stack<d_node>> cover_sets,
Stack<Stack<d_node>> new_cover_sets,
int current_scale, int max_scale) throws Exception {
new_cover_sets.clear();
for (; current_scale <= max_scale; current_scale++) {
d_node ele;
Stack<d_node> cover_set_currentscale = getCoverSet(current_scale,
cover_sets);
for (int i = 0; i < cover_set_currentscale.length; i++) { // ; ele != end;
// ele++) {
ele = cover_set_currentscale.element(i);
double upper_dist = new_upper_k.peek().distance + query_chi.max_dist
+ ele.n.max_dist;
if (shell(ele.dist, query_chi.parent_dist, upper_dist)) {
double d = Math.sqrt(m_DistanceFunction.distance(query_chi.p(), ele.n
.p(), upper_dist * upper_dist));
if (m_TreeStats != null)
m_TreeStats.incrPointCount();
if (d <= upper_dist) {
if (d < new_upper_k.peek().distance)
update(new_upper_k, d);
d_node temp = new d_node(d, ele.n);
new_cover_sets.element(current_scale).push(temp);
if (m_TreeStats != null)
m_TreeStats.incrIntNodeCount();
}// end if(d<=..
}// end if(shell(...
}// end for(coverset_i)
}// end for(scales)
}
/**
* Prints the given cover sets and zero set.
*
* @param cover_sets The cover sets to print.
* @param zero_set The zero set to print.
* @param current_scale The scale/level to start printing
* the cover sets from.
* @param max_scale The max scale/level to print the cover
* sets upto.
*/
void print_cover_sets(Stack<Stack<d_node>> cover_sets,
Stack<d_node> zero_set, int current_scale, int max_scale) {
d_node ele;
println("cover set = ");
for (; current_scale <= max_scale; current_scale++) {
println("" + current_scale);
for (int i = 0; i < cover_sets.element(current_scale).length; i++) {
ele = cover_sets.element(current_scale).element(i);
CoverTreeNode n = ele.n;
println(n.p());
}
}
println("infinity");
for (int i = 0; i < zero_set.length; i++) {
ele = zero_set.element(i);
CoverTreeNode n = ele.n;
println(n.p());
}
}
/**
* Swap two nodes in a cover set.
*
* @param a The index first node.
* @param b The index of second node.
* @param cover_set The cover set in which the two nodes are.
*/
protected void SWAP(int a, int b, Stack<d_node>cover_set) {
d_node tmp = cover_set.element(a);
cover_set.set(a, cover_set.element(b));
cover_set.set(b, tmp);
}
/**
* Returns the difference of two given nodes distance to
* the query. It is used in half-sorting a cover set.
*
* @param p1 The index of first node.
* @param p2 The index of second node.
* @param cover_set The cover set containing the two given
* nodes.
* @return dist_to_query_of_p1 - dist_to_query_of_p2
*/
protected double compare(final int p1, final int p2, Stack<d_node> cover_set) {
return cover_set.element(p1).dist - cover_set.element(p2).dist;
}
/**
* Half-sorts a cover set, so that nodes nearer to the query
* are at the front.
* @param cover_set The cover set to sort.
*/
protected void halfsort(Stack<d_node> cover_set) {
if(cover_set.length <= 1)
return;
int start=0;
int hi = cover_set.length-1;
int right = hi;
int left;
while (right > start) {
int mid = start + ((hi - start) >> 1);
boolean jumpover = false;
if (compare(mid, start, cover_set) < 0.0)
SWAP(mid, start, cover_set);
if (compare(hi, mid, cover_set) < 0.0)
SWAP(mid, hi, cover_set);
else
jumpover = true;
if (!jumpover && compare(mid, start, cover_set) < 0.0)
SWAP(mid, start, cover_set);
jump_over:
;
left = start + 1;
right = hi - 1;
do {
while (compare(left, mid, cover_set) < 0.0)
left++;
while (compare(mid, right, cover_set) < 0.0)
right--;
if (left < right) {
SWAP(left, right, cover_set);
if (mid == left)
mid = right;
else if (mid == right)
mid = left;
left++;
right--;
} else if (left == right) {
left++;
right--;
break;
}
} while (left <= right);
hi = right;
}
}
/**
* Function to check if a child node can be inside a query ball,
* without calculating the child node's distance to the query.
* This further avoids unnecessary distance calculation.
*
* @param parent_query_dist The distance of parent to the query
* @param child_parent_dist The distance of child to the parent.
* @param upper_bound The distance to the query of the best kth
* NN found so far.
* @return true If child can be inside the query ball.
*/
protected boolean shell(double parent_query_dist, double child_parent_dist, double upper_bound) {
return parent_query_dist - child_parent_dist <= upper_bound;
}
/**
* This functions adds nodes for inspection at the next level during NN
* search. The internal nodes are added to one of the cover sets (at
* the level of the child node which is added) and leaf nodes are
* added to the zero set.
*
* An optimization to consider:
* Make all distance evaluations occur in descend.
*
* Instead of passing a cover_set, pass a stack of cover sets. The
* last element holds d_nodes with your distance. The next lower
* element holds a d_node with the distance to your query parent,
* next = query grand parent, etc..
*
* Compute distances in the presence of the tighter upper bound.
* @param query The query (in shape of a cover tree node, as we
* are doing batch searching).
* @param upper_k Heap containing distances of best k-NNs found so
* far.
* @param current_scale The current scale/level being looked at in
* the tree.
* @param max_scale The max scale/level that has so far been looked
* at.
* @param cover_sets The cover sets of tree nodes for each level of
* our trees for.
* @param zero_set The set containing leaf nodes.
* @return A new max_scale, if we descend to a deeper level.
* @throws Exception If there is some problem (in updating the
* heap upper_k).
*/
protected int descend(final CoverTreeNode query, MyHeap upper_k,
int current_scale, int max_scale, // amk14comment: make sure this gets
// passed by reference in Java
Stack<Stack<d_node>> cover_sets, // amk14comment: contains children in
// set Q in paper
Stack<d_node> zero_set) // amk14comment: zeroset contains the children at
// the lowest level i.e. -infinity
throws Exception {
d_node parent;
Stack<d_node> cover_set_currentscale = getCoverSet(current_scale,
cover_sets);
for (int i = 0; i < cover_set_currentscale.length; i++) {
parent = cover_set_currentscale.element(i);
CoverTreeNode par = parent.n;
double upper_dist = upper_k.peek().distance + query.max_dist
+ query.max_dist; // *upper_bound + query->max_dist + query->max_dist;
if (parent.dist <= upper_dist + par.max_dist) {
CoverTreeNode chi;
if (par == m_Root && par.num_children == 0) // if our tree consists of
// only one root(which is
// also leaf) node
chi = par;
else
chi = par.children.element(0);
if (parent.dist <= upper_dist + chi.max_dist) { // amk14comment: looking
// at child_0 (which is
// the parent itself)
if (chi.num_children > 0) {
if (max_scale < chi.scale) {
max_scale = chi.scale;
}
d_node temp = new d_node(parent.dist, chi);
getCoverSet(chi.scale, cover_sets).push(temp);
if (m_TreeStats != null)
m_TreeStats.incrIntNodeCount();
} else if (parent.dist <= upper_dist) {
d_node temp = new d_node(parent.dist, chi);
zero_set.push(temp);
if (m_TreeStats != null)
m_TreeStats.incrLeafCount();
}
}
for (int c = 1; c < par.num_children; c++) {
chi = par.children.element(c);
double upper_chi = upper_k.peek().distance + chi.max_dist
+ query.max_dist + query.max_dist; // *upper_bound + chi.max_dist
// + query.max_dist +
// query.max_dist;
if (shell(parent.dist, chi.parent_dist, upper_chi)) { // amk14comment:parent_query_dist
// -
// child_parent_dist
// <= upper_chi - if child can be
// inside the shrunk query ball
// NOT the same as above parent->dist <= upper_dist + chi->max_dist
double d = Math.sqrt(m_DistanceFunction.distance(query.p(),
chi.p(), upper_chi * upper_chi, m_TreeStats));
if (m_TreeStats != null)
m_TreeStats.incrPointCount();
if (d <= upper_chi) { //if child is inside the shrunk query ball
if (d < upper_k.peek().distance) // *upper_bound)
update(upper_k, d);
if (chi.num_children > 0) {
if (max_scale < chi.scale) {
max_scale = chi.scale;
}
d_node temp = new d_node(d, chi);
getCoverSet(chi.scale, cover_sets).push(temp);
if (m_TreeStats != null)
m_TreeStats.incrIntNodeCount();
} else if (d <= upper_chi - chi.max_dist) {
d_node temp = new d_node(d, chi);
zero_set.push(temp);
if (m_TreeStats != null)
m_TreeStats.incrLeafCount();
}
}//end if(d<=upper_chi)
}//end if(shell(parent.dist,...
}//end for(child_1 to n)
}//end if(parent.dist<=upper_dist..
}//end for(covers_sets[current_scale][i])
return max_scale;
}
/**
* Does a brute force NN search on the nodes in the given zero set.
* A zero set might have some nodes added to it that were not k-NNs,
* so need to do a brute-force to pick only the k-NNs (without
* calculating distances, as each node in the zero set already had
* its distance calculated to the query, which is stored with the
* node).
*
* @param k The k in kNN.
* @param query The query.
* @param zero_set The zero set on which the brute force NN search
* is performed.
* @param upper_k The heap storing distances of k-NNs found during
* the search.
* @param results The returned k-NNs.
* @throws Exception If there is somem problem.
*/
protected void brute_nearest(final int k, final CoverTreeNode query,
Stack<d_node> zero_set, MyHeap upper_k, Stack<NeighborList> results)
throws Exception {
if (query.num_children > 0) {
Stack<d_node> new_zero_set = new Stack<d_node>();
CoverTreeNode query_chi = query.children.element(0);
brute_nearest(k, query_chi, zero_set, upper_k, results);
MyHeap new_upper_k = new MyHeap(k);
for (int i = 1; i < query.children.length; i++) {
query_chi = query.children.element(i);
setter(new_upper_k, upper_k.peek().distance + query_chi.parent_dist, k);
copy_zero_set(query_chi, new_upper_k, zero_set, new_zero_set);
brute_nearest(k, query_chi, new_zero_set, new_upper_k, results);
}
} else {
NeighborList temp = new NeighborList(k);
d_node ele;
for (int i = 0; i < zero_set.length; i++) {
ele = zero_set.element(i);
if (ele.dist <= upper_k.peek().distance) {
temp.insertSorted(ele.dist, ele.n.p()); // temp.push(ele.n.p());
}
}
results.push(temp);
}
}
/**
* Performs a recursive k-NN search for a given batch of queries provided in the
* form of a cover tree. P.S.: This function should not be called from outside.
* Outside classes should use kNearestNeighbours() instead.
*
* @param k The number of NNs to find.
* @param query_node The node of the query tree to start the search from.
* @param cover_sets The set of sets that contains internal
* nodes that were found to be inside the query ball at previous scales/levels
* (intially there would be just the root node at root level).
* @param zero_set The set that'll contain the leaf nodes that are found to
* be inside the query ball.
* @param current_scale The level/scale to do the search from (this value
* would be used to inspect the cover set in the provided set of cover sets).
* @param max_scale The max scale/level that has so far been inspected.
* @param upper_k The heap containing distances of the best k-NNs found so
* far (initialized to Double.POSITIVE_INFINITY).
* @param results The list of returned k-NNs.
* @throws Exception If there is some problem during the search.
*/
protected void internal_batch_nearest_neighbor(final int k,
final CoverTreeNode query_node,
Stack<Stack<d_node>> cover_sets,
Stack<d_node> zero_set,
int current_scale,
int max_scale,
MyHeap upper_k,
Stack<NeighborList> results) throws Exception {
if (current_scale > max_scale) { // All remaining points are in the zero set.
brute_nearest(k, query_node, zero_set, upper_k, results);
} else {
// Our query_node has too much scale. Reduce.
if (query_node.scale <= current_scale && query_node.scale != 100) { // amk14comment:if j>=i in paper
CoverTreeNode query_chi;
Stack<d_node> new_zero_set = new Stack<d_node>();
Stack<Stack<d_node>> new_cover_sets = new Stack<Stack<d_node>>();
MyHeap new_upper_k = new MyHeap(k);
for (int i = 1; i < query_node.num_children; i++) { //processing child_1 and onwards
query_chi = query_node.children.element(i);
setter(new_upper_k, upper_k.peek().distance + query_chi.parent_dist, k);
//copy the zero set that satisfy a certain bound to the new zero set
copy_zero_set(query_chi, new_upper_k, zero_set, new_zero_set);
//copy the coversets[current_scale] nodes that satisfy a certain
//bound to the new_cover_sets[current_scale]
copy_cover_sets(query_chi, new_upper_k, cover_sets, new_cover_sets,
current_scale, max_scale);
//search for the query_node child in the nodes nearer to it.
internal_batch_nearest_neighbor(k, query_chi, new_cover_sets,
new_zero_set, current_scale, max_scale, new_upper_k, results);
}
new_cover_sets = null;
new_zero_set = null;
new_upper_k = null;
// now doing child_0 //which is the parent itself, that's why we don't
// need new_zero_set or new_cover_sets
internal_batch_nearest_neighbor(k, query_node.children.element(0),
cover_sets, zero_set, current_scale, max_scale, upper_k, results);
} else { // reduce cover set scale -- amk14comment: if j<i in paper
Stack<d_node> cover_set_i = getCoverSet(current_scale, cover_sets);
// println("sorting");
halfsort(cover_set_i);
max_scale = descend(query_node, upper_k, current_scale, max_scale,
cover_sets, zero_set);
cover_set_i.clear();
current_scale++;
internal_batch_nearest_neighbor(k, query_node, cover_sets, zero_set,
current_scale, max_scale, upper_k, results);
}
}
}
/**
* Performs k-NN search for a batch of queries provided in the form
* of a cover tree. P.S.: Outside classes should call
* kNearestNeighbours().
*
* @param k The number of k-NNs to find.
* @param tree_root The root of the cover tree on which k-NN search
* is to be performed.
* @param query_root The root of the cover tree consisting of queries.
* @param results The list of returned k-NNs.
* @throws Exception If there is some problem during the search.
*/
protected void batch_nearest_neighbor(final int k, CoverTreeNode tree_root, CoverTreeNode query_root,
Stack<NeighborList> results) throws Exception {
//amk14comment: These contain the covering nodes at each level
Stack<Stack<d_node>> cover_sets = new Stack<Stack<d_node>>(100);
//amk14comment: These contain the nodes thought to be nearest at the leaf level
Stack<d_node> zero_set = new Stack<d_node>();
MyHeap upper_k = new MyHeap(k);
//probably not needed //amk14comment:initializes the array to MAXFLOAT
setter(upper_k, Double.POSITIVE_INFINITY, k);
// amk14comment:distance from top query point to top node point
double treeroot_to_query_dist = Math.sqrt(m_DistanceFunction.distance(
query_root.p(), tree_root.p(), Double.POSITIVE_INFINITY));
// amk14comment:probably stores the kth smallest distances encountered so
// far
update(upper_k, treeroot_to_query_dist);
d_node temp = new d_node(treeroot_to_query_dist, tree_root);
getCoverSet(0, cover_sets).push(temp);
// incrementing counts for the root node
if (m_TreeStats != null) {
m_TreeStats.incrPointCount();
if (tree_root.num_children > 0)
m_TreeStats.incrIntNodeCount();
else
m_TreeStats.incrLeafCount();
}
internal_batch_nearest_neighbor(k, query_root, cover_sets, zero_set, 0, 0,
upper_k, results);
}
/**
* Performs k-NN serach for a single given query/test Instance.
*
* @param target The query/test instance.
* @param k Number of k-NNs to find.
* @return List of k-NNs.
* @throws Exception If there is some problem during the search
* for k-NNs.
*/
protected NeighborList findKNearest(final Instance target, final int k) throws Exception {
Stack<d_node> cover_set_current = new Stack<d_node>(),
cover_set_next,
zero_set = new Stack<d_node>();
CoverTreeNode parent, child; d_node par;
MyHeap upper_k = new MyHeap(k);
double d = Math.sqrt(m_DistanceFunction.distance(m_Root.p(), target, Double.POSITIVE_INFINITY, m_TreeStats)),
upper_bound;
cover_set_current.push(new d_node(d, m_Root));
setter(upper_k, Double.POSITIVE_INFINITY, k);
this.update(upper_k, d);
//updating stats for the root node
if(m_TreeStats!=null) {
if(m_Root.num_children > 0)
m_TreeStats.incrIntNodeCount();
else
m_TreeStats.incrLeafCount();
m_TreeStats.incrPointCount();
}
//if root is the only node
if(m_Root.num_children==0) {
NeighborList list = new NeighborList(k);
list.insertSorted(d, m_Root.p());
return list;
}
//else
while(cover_set_current.length>0) {
cover_set_next = new Stack<d_node>();
for(int i=0; i<cover_set_current.length; i++) {
par = cover_set_current.element(i);
parent = par.n;
for(int c=0; c<parent.num_children; c++) {
child = parent.children.element(c);
upper_bound = upper_k.peek().distance;
if(c==0)
d = par.dist;
else {
d = upper_bound + child.max_dist;
d = Math.sqrt(m_DistanceFunction.distance(child.p(), target, d*d, m_TreeStats));
if(m_TreeStats!=null)
m_TreeStats.incrPointCount();
}
if(d <= (upper_bound + child.max_dist)) {
if(c>0 && d < upper_bound) {
update(upper_k, d);
}
if(child.num_children > 0) {
cover_set_next.push(new d_node(d, child));
if(m_TreeStats!=null)
m_TreeStats.incrIntNodeCount();
}
else if (d <= upper_bound){
zero_set.push(new d_node(d, child));
if(m_TreeStats!=null)
m_TreeStats.incrLeafCount();
}
}
} //end for current_set children
} //end for current_set elements
cover_set_current = cover_set_next;
} //end while(curret_set not empty)
NeighborList list = new NeighborList(k);
d_node tmpnode;
upper_bound = upper_k.peek().distance;
for(int i=0; i<zero_set.length; i++) {
tmpnode = zero_set.element(i);
if(tmpnode.dist <= upper_bound)
list.insertSorted(tmpnode.dist, tmpnode.n.p());
}
if(list.currentLength()<=0)
throw new Exception("Error: No neighbour found. This cannot happen");
return list;
}
/*********************************NNSearch related stuff above.********************/
/**
* Returns k-NNs of a given target instance, from among the previously
* supplied training instances (supplied through setInstances method)
* P.S.: May return more than k-NNs if more one instances have
* the same distance to the target as the kth NN.
*
* @param target The instance for which k-NNs are required.
* @param k The number of k-NNs to find.
* @return The k-NN instances of the given target instance.
* @throws Exception If there is some problem find the k-NNs.
*/
public Instances kNearestNeighbours(Instance target, int k) throws Exception {
if(m_Stats!=null)
m_Stats.searchStart();
CoverTree querytree = new CoverTree();
Instances insts = new Instances(m_Instances, 0);
insts.add(target);
querytree.setInstances(insts);
Stack<NeighborList> result = new Stack<NeighborList>();
batch_nearest_neighbor(k, this.m_Root, querytree.m_Root, result);
if(m_Stats!=null)
m_Stats.searchFinish();
insts = new Instances(m_Instances, 0);
NeighborNode node = result.element(0).getFirst();
m_DistanceList = new double[result.element(0).currentLength()];
int i=0;
while(node != null) {
insts.add(node.m_Instance);
m_DistanceList[i] = node.m_Distance;
i++; node = node.m_Next;
}
return insts;
}
/**
* Returns the NN instance of a given target instance, from among
* the previously supplied training instances.
*
* @param target The instance for which NN is required.
* @throws Exception If there is some problem finding the nearest
* neighbour.
* @return The NN instance of the target instance.
*/
public Instance nearestNeighbour(Instance target) throws Exception {
return kNearestNeighbours(target, 1).instance(0);
}
/**
* Returns the distances of the (k)-NN(s) found earlier
* by kNearestNeighbours()/nearestNeighbour().
*
* @throws Exception If the tree hasn't been built (by calling
* setInstances()), or none of kNearestNeighbours() or
* nearestNeighbour() has been called before.
* @return The distances (in the same order) of the k-NNs.
*/
public double[] getDistances() throws Exception {
if(m_Instances==null || m_DistanceList==null)
throw new Exception("The tree has not been supplied with a set of " +
"instances or getDistances() has been called " +
"before calling kNearestNeighbours().");
return m_DistanceList;
}
/**
* Checks if there is any instance with missing values. Throws an
* exception if there is, as KDTree does not handle missing values.
*
* @param instances the instances to check
* @throws Exception if missing values are encountered
*/
protected void checkMissing(Instances instances) throws Exception {
for (int i = 0; i < instances.numInstances(); i++) {
Instance ins = instances.instance(i);
for (int j = 0; j < ins.numValues(); j++) {
if (ins.index(j) != ins.classIndex())
if (ins.isMissingSparse(j)) {
throw new Exception("ERROR: KDTree can not deal with missing "
+ "values. Please run ReplaceMissingValues filter "
+ "on the dataset before passing it on to the KDTree.");
}
}
}
}
/**
* Builds the Cover Tree on the given set of instances.
*
* @param instances The insts on which the Cover Tree is to be
* built.
* @throws Exception If some error occurs while
* building the Cover Tree
*/
public void setInstances(Instances instances) throws Exception {
super.setInstances(instances);
buildCoverTree(instances);
}
/**
* Adds an instance to the cover tree.
* P.S.: The current version doesn't allow
* addition of instances after batch construction.
*
* @param ins The instance to add.
* @throws Exception Alway throws this, as current
* implementation doesn't allow addition of instances
* after building.
*/
public void update(Instance ins) throws Exception {
throw new Exception("BottomUpConstruction method does not allow addition " +
"of new Instances.");
}
/**
* Adds the given instance info. This implementation updates only the
* range datastructures of the EuclideanDistance. Nothing is
* required to be updated in the built Cover Tree.
*
* @param ins The instance to add the information of. Usually this is
* the test instance supplied to update the range of
* attributes in the distance function.
*/
public void addInstanceInfo(Instance ins) {
if(m_Instances!=null) {
try {
m_DistanceFunction.update(ins);
} catch(Exception ex) { ex.printStackTrace(); }
}
else
if(m_Instances==null)
throw new IllegalStateException("No instances supplied yet. Cannot update without"+
"supplying a set of instances first.");
}
/**
* Sets the distance function to use for nearest neighbour search.
* Currently only EuclideanDistance is supported.
*
* @param df the distance function to use
* @throws Exception if not EuclideanDistance
*/
public void setDistanceFunction(DistanceFunction df) throws Exception {
if (!(df instanceof EuclideanDistance))
throw new Exception("CoverTree currently only works with "
+ "EuclideanDistanceFunction.");
m_DistanceFunction = m_EuclideanDistance = (EuclideanDistance) df;
}
/**
* Returns the tip text for this property.
*
* @return tip text for this property suitable for
* displaying in the explorer/experimenter gui
*/
public String baseTipText() {
return "The base for the expansion constant.";
}
/**
* Returns the base in use for expansion constant.
*
* @return base currently in use.
*/
public double getBase() {
return m_Base;
}
/**
* Sets the base to use for expansion constant.
* The 2 in 2^i in the paper.
*
* @param b the new base;
*/
public void setBase(double b) {
m_Base = b;
}
/**
* Returns the size of the tree.
* (number of internal nodes + number of leaves)
*
* @return the size of the tree
*/
public double measureTreeSize() {
return m_NumNodes;
}
/**
* Returns the number of leaves.
*
* @return the number of leaves
*/
public double measureNumLeaves() {
return m_NumLeaves;
}
/**
* Returns the depth of the tree.
*
* @return the number of rules
*/
public double measureMaxDepth() {
return m_MaxDepth;
}
/**
* Returns an enumeration of the additional measure names.
*
* @return an enumeration of the measure names
*/
public Enumeration enumerateMeasures() {
Vector newVector = new Vector();
newVector.addElement("measureTreeSize");
newVector.addElement("measureNumLeaves");
newVector.addElement("measureMaxDepth");
if(m_Stats!=null) {
for(Enumeration e = m_Stats.enumerateMeasures(); e.hasMoreElements();) {
newVector.addElement(e.nextElement());
}
}
return newVector.elements();
}
/**
* Returns the value of the named measure.
*
* @param additionalMeasureName the name of the measure to query for
* its value
* @return the value of the named measure
* @throws IllegalArgumentException if the named measure is not supported
*/
public double getMeasure(String additionalMeasureName) {
if (additionalMeasureName.compareToIgnoreCase("measureMaxDepth") == 0) {
return measureMaxDepth();
} else if (additionalMeasureName.compareToIgnoreCase("measureTreeSize") == 0) {
return measureTreeSize();
} else if (additionalMeasureName.compareToIgnoreCase("measureNumLeaves") == 0) {
return measureNumLeaves();
} else if(m_Stats!=null) {
return m_Stats.getMeasure(additionalMeasureName);
} else {
throw new IllegalArgumentException(additionalMeasureName
+ " not supported (KDTree)");
}
}
/********Utility print functions.****** */
/**
* Prints a string to stdout.
*
* @param s The string to print.
*/
protected static void print(String s) {
System.out.print(s);
}
/**
* Prints a string to stdout followed by
* newline.
*
* @param s The string to print.
*/
protected static void println(String s) {
System.out.println(s);
}
/**
* Prints an object to stdout.
*
* @param o The object to print.
*/
protected static void print(Object o) {
System.out.print(o);
}
/**
* Prints an object to stdout followed by
* newline.
*
* @param o The object to print.
*/
protected static void println(Object o) {
System.out.println(o);
}
/**
* Prints the specified number of spaces.
*
* @param s The number of space characters to print.
*/
protected static void print_space(int s) {
for (int i = 0; i < s; i++)
System.out.print(" ");
}
/**
* Prints a cover tree starting from the given node.
*
* @param depth The depth of top_node.
* @param top_node The node to start printing from.
*/
protected static void print(int depth, CoverTreeNode top_node) {
print_space(depth);
println(top_node.p());
if (top_node.num_children > 0) {
print_space(depth);
print("scale = " + top_node.scale + "\n");
print_space(depth);
print("num children = " + top_node.num_children + "\n");
System.out.flush();
for (int i = 0; i < top_node.num_children; i++)
print(depth + 1, top_node.children.element(i)); // top_node.children[i]);
}
}
/**
* Returns the revision string.
*
* @return the revision
*/
public String getRevision() {
return RevisionUtils.extract("$Revision: 1.4 $");
}
/**
* Method for testing the class from command line.
*
* @param args The supplied command line arguments.
*/
public static void main(String[] args) {
if (args.length != 1) {
System.err.println("Usage: CoverTree <ARFF file>");
System.exit(-1);
}
try {
Instances insts = null;
if (args[0].endsWith(".csv")) {
CSVLoader csv = new CSVLoader();
csv.setFile(new File(args[0]));
insts = csv.getDataSet();
} else {
insts = new Instances(new BufferedReader(new FileReader(args[0])));
}
CoverTree tree = new CoverTree();
tree.setInstances(insts);
print("Created data tree:\n");
print(0, tree.m_Root);
println("");
} catch (Exception ex) {
ex.printStackTrace();
}
}
}