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
190
191
192
193
194
195
196
197
198
|
// -*- Mode: C++; tab-width: 2; -*-
// vi: set ts=2:
//
//
#include <BALL/QSAR/libsvmModel.h>
#define Malloc(type,n) (type *)malloc((n)*sizeof(type))
using namespace BALL::QSAR;
LibsvmModel::LibsvmModel(const QSARData& q, int k_type, double p1, double p2) : SVRModel(q, k_type, p1, p2)
{
type_="SVR";
svm_train_result_ = NULL;
use_nu_ = 1;
nu_ = 0.05;
p_ = 0.1;
eps_ = 1e-3;
C_ = 1;
createParameters();
x_space_ = NULL;
}
LibsvmModel::~LibsvmModel()
{
delete x_space_;
}
void LibsvmModel::train()
{
if (descriptor_matrix_.cols() == 0)
{
throw Exception::InconsistentUsage(__FILE__, __LINE__, "Data must be read into the model before training!");
}
kernel->calculateKernelMatrix(descriptor_matrix_, K_);
struct svm_problem* prob = NULL;
training_result_.resize(K_.rows(), Y_.cols());
offsets_.resize(Y_.cols());
for (int act = 0; act <= Y_.cols(); act++)
{
prob = createProblem(act);
svm_train_result_ = (LibsvmModel::svm_model*)svm_train(prob, ¶meters_);
free(prob->y);free(prob->x);free(prob);
const double* const *sv_coef = svm_train_result_->sv_coef;
//const svm_node* const *SV = svm_train_result_->SV;
for(int i=0;i<svm_train_result_->l;i++) // l=#support vectors == #compounds
{
for (int j = 0; j < svm_train_result_->nr_class-1; j++)
{
training_result_(i, j) = sv_coef[j][i];
}
}
offsets_(act) = svm_train_result_->rho[0];
//free(prob);
//free(prob->y); free(prob->x);
#ifdef LIBSVM_VERSION
svm_free_and_destroy_model((::svm_model**)&svm_train_result_);
#else
svm_destroy_model((::svm_model*)svm_train_result_);
#endif
}
}
struct svm_problem* LibsvmModel::createProblem(int response_id)
{
struct svm_problem* prob = Malloc(svm_problem, 1);
prob->l = K_.rows();
prob->y = Malloc(double, prob->l);
prob->x = Malloc(struct svm_node*, prob->l);
int elements = (K_.cols()+2)*K_.rows();
free(x_space_);
x_space_ = Malloc(struct svm_node, elements);
int cols = K_.cols();
int index = 0;
for (int i = 0; i < K_.rows(); i++)
{
//prob->x[i-1] = Malloc(struct svm_node, cols+1);
prob->x[i] = &x_space_[index];
prob->y[i] = Y_(i, response_id);
x_space_[index].index = 0;
x_space_[index].value = i; // numer of current row
index++;
for (int j = 0; j < cols; j++)
{
//prob->x[i-1][j-1].index = j;
//prob->x[i-1][j-1].value = K_(i, j);
x_space_[index].index = j;
x_space_[index].value = K_(i, j);
index++;
}
//prob->x[i-1][cols].index = -1;
//prob->x[i-1][cols].value = '?';
x_space_[index].index = -1;
x_space_[index].value = '?';
index++;
}
return prob;
}
void LibsvmModel::createParameters()
{
parameters_.kernel_type = 4; // use precomputed kernel !
parameters_.cache_size = 100;
if (use_nu_)
{
parameters_.svm_type = 4;
}
else
{
parameters_.svm_type = 3;
}
parameters_.nu = nu_;
parameters_.C = C_;
parameters_.eps = eps_;
parameters_.p = p_;
parameters_.shrinking = use_shrinking_;
parameters_.probability = 0;
}
// RowVector LibsvmModel::predict(const Eigen::VectorXd& substance, bool transform)
// {
// if(svm_train_result_==NULL)
// {
// throw Exception::InconsistentUsage(__FILE__,__LINE__,"Model must be trained before it can predict the activitiy of substances!");
// }
// RowVector input=getSubstanceVector(substance, transform);
// Matrix K_t(input.rows(), descriptor_matrix_.rows());
// kernel->calculateKernelMatrix(K_,input, descriptor_matrix_, K_t);
//
// svm_node* node = Malloc(struct svm_node, K_t.cols()+1);
//
// for(int i=1; i<=K_t.cols(); i++)
// {
// node[i-1].index = i-1;
// node[i-1].value = K_t(1,i);
// }
// node[K_t.cols()].index = -1;
// node[K_t.cols()].value = '?';
//
// double res = svm_predict(svm_train_result_, node);
//
// RowVector rv(1);
// rv << res;
//
// if(transform && y_transformations_.cols()!=0)
// {
// backTransformPrediction(rv);
// }
//
// free(node);
// return rv;
// }
void LibsvmModel::setParameters(vector<double>& v)
{
if (v.size() != 6)
{
String c = "Wrong number of model parameters! Needed: 6;";
c = c+" given: "+String(v.size());
throw Exception::ModelParameterError(__FILE__, __LINE__, c.c_str());
}
String d = v[0];
use_nu_ = d.toBool();
d = v[1];
use_shrinking_ = d.toBool();
nu_ = v[2];
p_ = v[3];
eps_ = v[4];
C_ = v[5];
createParameters();
}
vector<double> LibsvmModel::getParameters() const
{
vector<double> d;
d.push_back(use_nu_);
d.push_back(use_shrinking_);
d.push_back(nu_);
d.push_back(p_);
d.push_back(eps_);
d.push_back(C_);
return d;
}
|
