1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
|
void svm(cv::Mat& trainingData, cv::Mat& trainingClasses, cv::Mat& testData, cv::Mat& testClasses, int size)
{
CvSVMParams param = CvSVMParams();
param.svm_type = CvSVM::C_SVC;
param.kernel_type = CvSVM::RBF; //CvSVM::RBF, CvSVM::LINEAR ... the best choice
param.degree = 0; // for poly
param.gamma = 20; // for poly/rbf/sigmoid
param.coef0 = 0; // for poly/sigmoid
/*Parameters in the generalized SVM optimization problem */
param.C = 7; // for CV_SVM_C_SVC, CV_SVM_EPS_SVR and CV_SVM_NU_SVR
param.nu = 1.0; // for CV_SVM_NU_SVC, CV_SVM_ONE_CLASS, and CV_SVM_NU_SVR
param.p = 0.0; // for CV_SVM_EPS_SVR
param.class_weights = NULL; // for CV_SVM_C_SVC
param.term_crit.type = CV_TERMCRIT_ITER +CV_TERMCRIT_EPS;
param.term_crit.max_iter = 1000; // maximum number of iteration in training
param.term_crit.epsilon = 1e-6; // epsilon is the error to stop training
/* SVM training (use train auto for OpenCV>=2.0)
The method trains the SVM model. It follows the conventions of the generic train “method” with the following limitations: only the CV _ ROW _ SAMPLE data layout is supported,
the input variables are all ordered, the output variables can be either categorical ( _params.svm_type=CvSVM::C_SVC or _params.svm_type=CvSVM::NU_SVC ),
or ordered ( _params.svm_type=CvSVM::EPS_SVR or _params.svm_type=CvSVM::NU_SVR ), or not required at all ( _params.svm_type=CvSVM::ONE_CLASS ), missing measurements are not supported.
All the other parameters are gathered in CvSVMParams structure.
*/
CvSVM SVM;
SVM.train_auto(trainingData, trainingClasses, cv::Mat(), cv::Mat(), param); // PLANTAGE
// Mat predicted(testClasses.rows, 1, CV_32F);
/*
for(int i = 0; i < testData.rows; i++)
{
Mat sample = testData.row(i);
//float x = sample.at<float>(0,0);
//float y = sample.at<float>(0,1);
float result = svm.predict(sample);
cout << "result= " << result << "\n";
}
*/
/*
cout << "Accuracy_{SVM} = " << evaluate(predicted, testClasses) << endl;
plot_binary(testData, predicted, "Predictions SVM", size);
// plot support vectors
if(plotSupportVectors)
{
Mat plot_sv(size, size, CV_8UC3);
plot_sv.setTo(cv::Scalar(255.0, 255.0, 255.0));
int svec_count = svm.get_support_vector_count();
for(int vecNum = 0; vecNum < svec_count; vecNum++)
{
const float* vec = svm.get_support_vector(vecNum);
circle(plot_sv, Point(vec[0]*size, vec[1]*size), 3 , CV_RGB(0, 0, 0));
}
imshow("Support Vectors", plot_sv);
}
*/
}
int main(int argc, char** argv )
{
if (argc == 6)
{
//argv[1] = file: ad_cranfield.data
//argv[2] = line1_train;
//argv[3] = line2_train;
//argv[4] = line1_test;
//argv[5] = line2_test;
FILE* frandom = NULL;
char* line1_train = argv[2];
char* line2_train = argv[3];
char* line1_test = argv[4];
char* line2_test = argv[5];
// get min/max line numbers
int min_line_train = min(atoi(line1_train), atoi(line2_train)); // atoi convert string to integer, ex: "5" => 5
int max_line_train = max(atoi(line1_train), atoi(line2_train));
int min_line_test = min(atoi(line1_test), atoi(line2_test));
int max_line_test = max(atoi(line1_test), atoi(line2_test));
int NUMBER_OF_TRAINING_SAMPLES = max_line_train - min_line_train;
int NUMBER_OF_TESTING_SAMPLES = max_line_test - min_line_test;
cout << "NUMBER_OF_TRAINING_SAMPLES = " << NUMBER_OF_TRAINING_SAMPLES << "\n";
cout << "NUMBER_OF_TESTING_SAMPLES = " << NUMBER_OF_TESTING_SAMPLES << "\n";
FILE* f = fopen( argv[1], "r" );
if( !f )
{
printf("ERROR: cannot read file \n");
return 0; // all not OK
}
FILE* ftrain = f;
FILE* ftest = f;
frandom = f;
randomize (f, frandom, "frandom");
selectlines(line1_train, line2_train, frandom, ftrain, "fww");
selectlines(line1_test, line2_test, frandom, ftest, "fww");
// define training data storage matrices (one for attribute examples, one for classifications)
Mat training_data = Mat(NUMBER_OF_TRAINING_SAMPLES, ATTRIBUTES_PER_SAMPLE, CV_32FC1); // CV_32FC1 constant which assumes float type and train and res matrices just reinterpret bits of your int and doubles as floats.
Mat training_classifications = Mat(NUMBER_OF_TRAINING_SAMPLES, 1, CV_32FC1); // 1.0 = class 1 = ad and 0.0 = class 2 = nonad
//define testing data storage matrices
Mat testing_data = Mat(NUMBER_OF_TESTING_SAMPLES, ATTRIBUTES_PER_SAMPLE, CV_32FC1);
Mat testing_classifications = Mat(NUMBER_OF_TESTING_SAMPLES, 1, CV_32FC1); // 1.0 = class 1 = ad and 0.0 = class 2 = nonad
// define all the attributes as numerical (** not needed for all ML techniques **)
Mat var_type = Mat(ATTRIBUTES_PER_SAMPLE + 1, 1, CV_8U );
var_type.setTo(Scalar(CV_VAR_NUMERICAL)); // all inputs are numerical
// this is a classification problem so reset the last (+1) output
// var_type element to CV_VAR_CATEGORICAL (** not needed for all ML techniques **)
var_type.at<uchar>(ATTRIBUTES_PER_SAMPLE, 0) = CV_VAR_CATEGORICAL;
// load training and testing data sets
if (read_data_from_csv(ftrain, training_data, training_classifications, NUMBER_OF_TRAINING_SAMPLES) && read_data_from_csv(ftest, testing_data, testing_classifications, NUMBER_OF_TESTING_SAMPLES))
{
//*********************************************************************
// DO YOUR ML HERE
//*********************************************************************
// all matrix memory freed by destructors
// all OK : main returns 0
svm(training_data, training_classifications, testing_data, testing_classifications, 100 );
return 0;
}
else
{
// not OK : main returns -1
return -1;
}
}
else
{
return -1;
}
}
/******************************************************************************/ |
