2009-02-17 09:45:57 +08:00
|
|
|
// The contents of this file are in the public domain. See LICENSE_FOR_EXAMPLE_PROGRAMS.txt
|
2008-05-31 06:52:51 +08:00
|
|
|
/*
|
|
|
|
This is an example illustrating the use of the kkmeans object
|
|
|
|
from the dlib C++ Library.
|
|
|
|
|
|
|
|
The kkmeans object is an implementation of a kernelized k-means clustering
|
|
|
|
algorithm. It is implemented by using the kcentroid object to represent
|
|
|
|
each center found by the usual k-means clustering algorithm.
|
|
|
|
|
|
|
|
So this object allows you to perform non-linear clustering in the same way
|
|
|
|
a svm classifier finds non-linear decision surfaces.
|
|
|
|
|
|
|
|
This example will make points from 3 classes and perform kernelized k-means
|
|
|
|
clustering on those points.
|
|
|
|
|
|
|
|
The classes are as follows:
|
|
|
|
- points very close to the origin
|
|
|
|
- points on the circle of radius 10 around the origin
|
|
|
|
- points that are on a circle of radius 4 but not around the origin at all
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include <iostream>
|
|
|
|
#include <vector>
|
|
|
|
|
|
|
|
#include "dlib/svm.h"
|
|
|
|
#include "dlib/rand.h"
|
|
|
|
|
|
|
|
using namespace std;
|
|
|
|
using namespace dlib;
|
|
|
|
|
|
|
|
int main()
|
|
|
|
{
|
|
|
|
// Here we declare that our samples will be 2 dimensional column vectors.
|
2008-10-09 07:42:24 +08:00
|
|
|
// (Note that if you don't know the dimensionality of your vectors at compile time
|
|
|
|
// you can change the 2 to a 0 and then set the size at runtime)
|
2008-05-31 06:52:51 +08:00
|
|
|
typedef matrix<double,2,1> sample_type;
|
|
|
|
|
|
|
|
// Now we are making a typedef for the kind of kernel we want to use. I picked the
|
|
|
|
// radial basis kernel because it only has one parameter and generally gives good
|
|
|
|
// results without much fiddling.
|
|
|
|
typedef radial_basis_kernel<sample_type> kernel_type;
|
|
|
|
|
2008-09-15 00:58:51 +08:00
|
|
|
|
2009-03-16 07:24:08 +08:00
|
|
|
// Here we declare an instance of the kcentroid object. It is the object used to
|
2009-03-16 11:26:54 +08:00
|
|
|
// represent each of the centers used for clustering. The kcentroid has 3 parameters
|
2009-03-16 07:24:08 +08:00
|
|
|
// you need to set. The first argument to the constructor is the kernel we wish to
|
|
|
|
// use. The second is a parameter that determines the numerical accuracy with which
|
|
|
|
// the object will perform part of the learning algorithm. Generally, smaller values
|
|
|
|
// give better results but cause the algorithm to attempt to use more support vectors
|
|
|
|
// (and thus run slower and use more memory). The third argument, however, is the
|
|
|
|
// maximum number of support vectors a kcentroid is allowed to use. So you can use
|
2009-03-16 11:26:54 +08:00
|
|
|
// it to control the runtime complexity.
|
|
|
|
kcentroid<kernel_type> kc(kernel_type(0.1),0.01, 8);
|
2008-05-31 06:52:51 +08:00
|
|
|
|
|
|
|
// Now we make an instance of the kkmeans object and tell it to use kcentroid objects
|
|
|
|
// that are configured with the parameters from the kc object we defined above.
|
|
|
|
kkmeans<kernel_type> test(kc);
|
|
|
|
|
|
|
|
std::vector<sample_type> samples;
|
|
|
|
std::vector<sample_type> initial_centers;
|
|
|
|
|
|
|
|
sample_type m;
|
|
|
|
|
|
|
|
dlib::rand::float_1a rnd;
|
|
|
|
|
2008-06-17 08:05:08 +08:00
|
|
|
// we will make 50 points from each class
|
|
|
|
const long num = 50;
|
2008-05-31 06:52:51 +08:00
|
|
|
|
|
|
|
// make some samples near the origin
|
|
|
|
double radius = 0.5;
|
|
|
|
for (long i = 0; i < num; ++i)
|
|
|
|
{
|
2008-06-17 08:05:08 +08:00
|
|
|
double sign = 1;
|
|
|
|
if (rnd.get_random_double() < 0.5)
|
|
|
|
sign = -1;
|
2008-05-31 06:52:51 +08:00
|
|
|
m(0) = 2*radius*rnd.get_random_double()-radius;
|
2008-06-17 08:05:08 +08:00
|
|
|
m(1) = sign*sqrt(radius*radius - m(0)*m(0));
|
2008-05-31 06:52:51 +08:00
|
|
|
|
|
|
|
// add this sample to our set of samples we will run k-means
|
|
|
|
samples.push_back(m);
|
|
|
|
}
|
|
|
|
|
|
|
|
// make some samples in a circle around the origin but far away
|
|
|
|
radius = 10.0;
|
|
|
|
for (long i = 0; i < num; ++i)
|
|
|
|
{
|
2008-06-17 08:05:08 +08:00
|
|
|
double sign = 1;
|
|
|
|
if (rnd.get_random_double() < 0.5)
|
|
|
|
sign = -1;
|
2008-05-31 06:52:51 +08:00
|
|
|
m(0) = 2*radius*rnd.get_random_double()-radius;
|
2008-06-17 08:05:08 +08:00
|
|
|
m(1) = sign*sqrt(radius*radius - m(0)*m(0));
|
2008-05-31 06:52:51 +08:00
|
|
|
|
|
|
|
// add this sample to our set of samples we will run k-means
|
|
|
|
samples.push_back(m);
|
|
|
|
}
|
|
|
|
|
2008-06-17 08:05:08 +08:00
|
|
|
// make some samples in a circle around the point (25,25)
|
2008-05-31 06:52:51 +08:00
|
|
|
radius = 4.0;
|
|
|
|
for (long i = 0; i < num; ++i)
|
|
|
|
{
|
2008-06-17 08:05:08 +08:00
|
|
|
double sign = 1;
|
|
|
|
if (rnd.get_random_double() < 0.5)
|
|
|
|
sign = -1;
|
2008-05-31 06:52:51 +08:00
|
|
|
m(0) = 2*radius*rnd.get_random_double()-radius;
|
2008-06-17 08:05:08 +08:00
|
|
|
m(1) = sign*sqrt(radius*radius - m(0)*m(0));
|
2008-05-31 06:52:51 +08:00
|
|
|
|
|
|
|
// translate this point away from the origin
|
|
|
|
m(0) += 25;
|
|
|
|
m(1) += 25;
|
|
|
|
|
|
|
|
// add this sample to our set of samples we will run k-means
|
|
|
|
samples.push_back(m);
|
|
|
|
}
|
|
|
|
|
|
|
|
// tell the kkmeans object we made that we want to run k-means with k set to 3.
|
|
|
|
// (i.e. we want 3 clusters)
|
|
|
|
test.set_number_of_centers(3);
|
|
|
|
|
|
|
|
// You need to pick some initial centers for the k-means algorithm. So here
|
2008-06-15 23:16:26 +08:00
|
|
|
// we will use the dlib::pick_initial_centers() function which tries to find
|
|
|
|
// n points that are far apart (basically).
|
|
|
|
pick_initial_centers(3, initial_centers, samples, test.get_kernel());
|
2008-05-31 06:52:51 +08:00
|
|
|
|
2008-07-06 01:18:12 +08:00
|
|
|
// now run the k-means algorithm on our set of samples.
|
|
|
|
test.train(samples,initial_centers);
|
2008-05-31 06:52:51 +08:00
|
|
|
|
|
|
|
// now loop over all our samples and print out their predicted class. In this example
|
|
|
|
// all points are correctly identified.
|
|
|
|
for (unsigned long i = 0; i < samples.size()/3; ++i)
|
|
|
|
{
|
|
|
|
cout << test(samples[i]) << " ";
|
|
|
|
cout << test(samples[i+num]) << " ";
|
|
|
|
cout << test(samples[i+2*num]) << "\n";
|
|
|
|
}
|
|
|
|
|
2009-03-16 07:24:08 +08:00
|
|
|
// Now print out how many support vectors each center used. Note that
|
|
|
|
// the maximum number of 8 was reached. If you went back to the kcentroid
|
|
|
|
// constructor and changed the 8 to some bigger number you would see that these
|
|
|
|
// numbers would go up. However, 8 is all we need to correctly cluster this dataset.
|
|
|
|
cout << "num sv for center 0: " << test.get_kcentroid(0).dictionary_size() << endl;
|
|
|
|
cout << "num sv for center 1: " << test.get_kcentroid(1).dictionary_size() << endl;
|
|
|
|
cout << "num sv for center 2: " << test.get_kcentroid(2).dictionary_size() << endl;
|
|
|
|
|
2008-05-31 06:52:51 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
|