MCPcopy Create free account
hub / github.com/davisking/dlib / main

Function main

examples/svm_pegasos_ex.cpp:28–159  ·  view source on GitHub ↗

Source from the content-addressed store, hash-verified

26
27
28int main()
29{
30 // The svm functions use column vectors to contain a lot of the data on which they
31 // operate. So the first thing we do here is declare a convenient typedef.
32
33 // This typedef declares a matrix with 2 rows and 1 column. It will be the
34 // object that contains each of our 2 dimensional samples. (Note that if you wanted
35 // more than 2 features in this vector you can simply change the 2 to something else.
36 // Or if you don't know how many features you want until runtime then you can put a 0
37 // here and use the matrix.set_size() member function)
38 typedef matrix<double, 2, 1> sample_type;
39
40
41 // This is a typedef for the type of kernel we are going to use in this example.
42 // In this case I have selected the radial basis kernel that can operate on our
43 // 2D sample_type objects
44 typedef radial_basis_kernel<sample_type> kernel_type;
45
46
47 // Here we create an instance of the pegasos svm trainer object we will be using.
48 svm_pegasos<kernel_type> trainer;
49 // Here we setup the parameters to this object. See the dlib documentation for a
50 // description of what these parameters are.
51 trainer.set_lambda(0.00001);
52 trainer.set_kernel(kernel_type(0.005));
53
54 // Set the maximum number of support vectors we want the trainer object to use
55 // in representing the decision function it is going to learn. In general,
56 // supplying a bigger number here will only ever give you a more accurate
57 // answer. However, giving a smaller number will make the algorithm run
58 // faster and decision rules that involve fewer support vectors also take
59 // less time to evaluate.
60 trainer.set_max_num_sv(10);
61
62 std::vector<sample_type> samples;
63 std::vector<double> labels;
64
65 // make an instance of a sample matrix so we can use it below
66 sample_type sample, center;
67
68 center = 20, 20;
69
70 // Now let's go into a loop and randomly generate 1000 samples.
71 srand(time(0));
72 for (int i = 0; i < 10000; ++i)
73 {
74 // Make a random sample vector.
75 sample = randm(2,1)*40 - center;
76
77 // Now if that random vector is less than 10 units from the origin then it is in
78 // the +1 class.
79 if (length(sample) <= 10)
80 {
81 // let the svm_pegasos learn about this sample
82 trainer.train(sample,+1);
83
84 // save this sample so we can use it with the batch training examples below
85 samples.push_back(sample);

Callers

nothing calls this directly

Calls 11

cross_validate_trainerFunction · 0.85
set_max_num_svMethod · 0.80
randmFunction · 0.50
lengthFunction · 0.50
batch_cachedFunction · 0.50
verbose_batch_cachedFunction · 0.50
set_lambdaMethod · 0.45
set_kernelMethod · 0.45
trainMethod · 0.45
push_backMethod · 0.45
clearMethod · 0.45

Tested by

no test coverage detected