liblinear-ruby 1.0.1 → 1.0.2
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/README.md +1 -1
- data/ext/blasp.h +8 -8
- data/ext/daxpy.c +3 -3
- data/ext/ddot.c +3 -3
- data/ext/dnrm2.c +7 -7
- data/ext/dscal.c +4 -4
- data/ext/liblinear_wrap.cxx +382 -382
- data/ext/linear.cpp +44 -55
- data/ext/linear.h +5 -1
- data/ext/tron.cpp +13 -5
- data/ext/tron.h +1 -1
- data/lib/liblinear.rb +2 -0
- data/lib/liblinear/version.rb +1 -1
- metadata +2 -41
- data/liblinear-2.1/COPYRIGHT +0 -31
- data/liblinear-2.1/Makefile +0 -37
- data/liblinear-2.1/Makefile.win +0 -24
- data/liblinear-2.1/README +0 -600
- data/liblinear-2.1/blas/Makefile +0 -22
- data/liblinear-2.1/blas/blas.h +0 -25
- data/liblinear-2.1/blas/blasp.h +0 -438
- data/liblinear-2.1/blas/daxpy.c +0 -57
- data/liblinear-2.1/blas/ddot.c +0 -58
- data/liblinear-2.1/blas/dnrm2.c +0 -70
- data/liblinear-2.1/blas/dscal.c +0 -52
- data/liblinear-2.1/heart_scale +0 -270
- data/liblinear-2.1/linear.cpp +0 -3053
- data/liblinear-2.1/linear.def +0 -22
- data/liblinear-2.1/linear.h +0 -79
- data/liblinear-2.1/matlab/Makefile +0 -49
- data/liblinear-2.1/matlab/README +0 -208
- data/liblinear-2.1/matlab/libsvmread.c +0 -212
- data/liblinear-2.1/matlab/libsvmwrite.c +0 -119
- data/liblinear-2.1/matlab/linear_model_matlab.c +0 -176
- data/liblinear-2.1/matlab/linear_model_matlab.h +0 -2
- data/liblinear-2.1/matlab/make.m +0 -22
- data/liblinear-2.1/matlab/predict.c +0 -341
- data/liblinear-2.1/matlab/train.c +0 -492
- data/liblinear-2.1/predict.c +0 -243
- data/liblinear-2.1/python/Makefile +0 -4
- data/liblinear-2.1/python/README +0 -380
- data/liblinear-2.1/python/liblinear.py +0 -323
- data/liblinear-2.1/python/liblinearutil.py +0 -270
- data/liblinear-2.1/train.c +0 -449
- data/liblinear-2.1/tron.cpp +0 -241
- data/liblinear-2.1/tron.h +0 -35
- data/liblinear-2.1/windows/liblinear.dll +0 -0
- data/liblinear-2.1/windows/libsvmread.mexw64 +0 -0
- data/liblinear-2.1/windows/libsvmwrite.mexw64 +0 -0
- data/liblinear-2.1/windows/predict.exe +0 -0
- data/liblinear-2.1/windows/predict.mexw64 +0 -0
- data/liblinear-2.1/windows/train.exe +0 -0
- data/liblinear-2.1/windows/train.mexw64 +0 -0
@@ -1,492 +0,0 @@
|
|
1
|
-
#include <math.h>
|
2
|
-
#include <stdlib.h>
|
3
|
-
#include <string.h>
|
4
|
-
#include <ctype.h>
|
5
|
-
#include "linear.h"
|
6
|
-
|
7
|
-
#include "mex.h"
|
8
|
-
#include "linear_model_matlab.h"
|
9
|
-
|
10
|
-
#ifdef MX_API_VER
|
11
|
-
#if MX_API_VER < 0x07030000
|
12
|
-
typedef int mwIndex;
|
13
|
-
#endif
|
14
|
-
#endif
|
15
|
-
|
16
|
-
#define CMD_LEN 2048
|
17
|
-
#define Malloc(type,n) (type *)malloc((n)*sizeof(type))
|
18
|
-
#define INF HUGE_VAL
|
19
|
-
|
20
|
-
void print_null(const char *s) {}
|
21
|
-
void print_string_matlab(const char *s) {mexPrintf(s);}
|
22
|
-
|
23
|
-
void exit_with_help()
|
24
|
-
{
|
25
|
-
mexPrintf(
|
26
|
-
"Usage: model = train(training_label_vector, training_instance_matrix, 'liblinear_options', 'col');\n"
|
27
|
-
"liblinear_options:\n"
|
28
|
-
"-s type : set type of solver (default 1)\n"
|
29
|
-
" for multi-class classification\n"
|
30
|
-
" 0 -- L2-regularized logistic regression (primal)\n"
|
31
|
-
" 1 -- L2-regularized L2-loss support vector classification (dual)\n"
|
32
|
-
" 2 -- L2-regularized L2-loss support vector classification (primal)\n"
|
33
|
-
" 3 -- L2-regularized L1-loss support vector classification (dual)\n"
|
34
|
-
" 4 -- support vector classification by Crammer and Singer\n"
|
35
|
-
" 5 -- L1-regularized L2-loss support vector classification\n"
|
36
|
-
" 6 -- L1-regularized logistic regression\n"
|
37
|
-
" 7 -- L2-regularized logistic regression (dual)\n"
|
38
|
-
" for regression\n"
|
39
|
-
" 11 -- L2-regularized L2-loss support vector regression (primal)\n"
|
40
|
-
" 12 -- L2-regularized L2-loss support vector regression (dual)\n"
|
41
|
-
" 13 -- L2-regularized L1-loss support vector regression (dual)\n"
|
42
|
-
"-c cost : set the parameter C (default 1)\n"
|
43
|
-
"-p epsilon : set the epsilon in loss function of SVR (default 0.1)\n"
|
44
|
-
"-e epsilon : set tolerance of termination criterion\n"
|
45
|
-
" -s 0 and 2\n"
|
46
|
-
" |f'(w)|_2 <= eps*min(pos,neg)/l*|f'(w0)|_2,\n"
|
47
|
-
" where f is the primal function and pos/neg are # of\n"
|
48
|
-
" positive/negative data (default 0.01)\n"
|
49
|
-
" -s 11\n"
|
50
|
-
" |f'(w)|_2 <= eps*|f'(w0)|_2 (default 0.001)\n"
|
51
|
-
" -s 1, 3, 4 and 7\n"
|
52
|
-
" Dual maximal violation <= eps; similar to libsvm (default 0.1)\n"
|
53
|
-
" -s 5 and 6\n"
|
54
|
-
" |f'(w)|_1 <= eps*min(pos,neg)/l*|f'(w0)|_1,\n"
|
55
|
-
" where f is the primal function (default 0.01)\n"
|
56
|
-
" -s 12 and 13\n"
|
57
|
-
" |f'(alpha)|_1 <= eps |f'(alpha0)|,\n"
|
58
|
-
" where f is the dual function (default 0.1)\n"
|
59
|
-
"-B bias : if bias >= 0, instance x becomes [x; bias]; if < 0, no bias term added (default -1)\n"
|
60
|
-
"-wi weight: weights adjust the parameter C of different classes (see README for details)\n"
|
61
|
-
"-v n: n-fold cross validation mode\n"
|
62
|
-
"-C : find parameter C (only for -s 0 and 2)\n"
|
63
|
-
"-q : quiet mode (no outputs)\n"
|
64
|
-
"col:\n"
|
65
|
-
" if 'col' is setted, training_instance_matrix is parsed in column format, otherwise is in row format\n"
|
66
|
-
);
|
67
|
-
}
|
68
|
-
|
69
|
-
// liblinear arguments
|
70
|
-
struct parameter param; // set by parse_command_line
|
71
|
-
struct problem prob; // set by read_problem
|
72
|
-
struct model *model_;
|
73
|
-
struct feature_node *x_space;
|
74
|
-
int flag_cross_validation;
|
75
|
-
int flag_find_C;
|
76
|
-
int flag_C_specified;
|
77
|
-
int flag_solver_specified;
|
78
|
-
int col_format_flag;
|
79
|
-
int nr_fold;
|
80
|
-
double bias;
|
81
|
-
|
82
|
-
|
83
|
-
void do_find_parameter_C(double *best_C, double *best_rate)
|
84
|
-
{
|
85
|
-
double start_C;
|
86
|
-
double max_C = 1024;
|
87
|
-
if (flag_C_specified)
|
88
|
-
start_C = param.C;
|
89
|
-
else
|
90
|
-
start_C = -1.0;
|
91
|
-
find_parameter_C(&prob, ¶m, nr_fold, start_C, max_C, best_C, best_rate);
|
92
|
-
mexPrintf("Best C = %lf CV accuracy = %g%%\n", *best_C, 100.0**best_rate);
|
93
|
-
}
|
94
|
-
|
95
|
-
|
96
|
-
double do_cross_validation()
|
97
|
-
{
|
98
|
-
int i;
|
99
|
-
int total_correct = 0;
|
100
|
-
double total_error = 0;
|
101
|
-
double sumv = 0, sumy = 0, sumvv = 0, sumyy = 0, sumvy = 0;
|
102
|
-
double *target = Malloc(double, prob.l);
|
103
|
-
double retval = 0.0;
|
104
|
-
|
105
|
-
cross_validation(&prob,¶m,nr_fold,target);
|
106
|
-
if(param.solver_type == L2R_L2LOSS_SVR ||
|
107
|
-
param.solver_type == L2R_L1LOSS_SVR_DUAL ||
|
108
|
-
param.solver_type == L2R_L2LOSS_SVR_DUAL)
|
109
|
-
{
|
110
|
-
for(i=0;i<prob.l;i++)
|
111
|
-
{
|
112
|
-
double y = prob.y[i];
|
113
|
-
double v = target[i];
|
114
|
-
total_error += (v-y)*(v-y);
|
115
|
-
sumv += v;
|
116
|
-
sumy += y;
|
117
|
-
sumvv += v*v;
|
118
|
-
sumyy += y*y;
|
119
|
-
sumvy += v*y;
|
120
|
-
}
|
121
|
-
mexPrintf("Cross Validation Mean squared error = %g\n",total_error/prob.l);
|
122
|
-
mexPrintf("Cross Validation Squared correlation coefficient = %g\n",
|
123
|
-
((prob.l*sumvy-sumv*sumy)*(prob.l*sumvy-sumv*sumy))/
|
124
|
-
((prob.l*sumvv-sumv*sumv)*(prob.l*sumyy-sumy*sumy))
|
125
|
-
);
|
126
|
-
retval = total_error/prob.l;
|
127
|
-
}
|
128
|
-
else
|
129
|
-
{
|
130
|
-
for(i=0;i<prob.l;i++)
|
131
|
-
if(target[i] == prob.y[i])
|
132
|
-
++total_correct;
|
133
|
-
mexPrintf("Cross Validation Accuracy = %g%%\n",100.0*total_correct/prob.l);
|
134
|
-
retval = 100.0*total_correct/prob.l;
|
135
|
-
}
|
136
|
-
|
137
|
-
free(target);
|
138
|
-
return retval;
|
139
|
-
}
|
140
|
-
|
141
|
-
// nrhs should be 3
|
142
|
-
int parse_command_line(int nrhs, const mxArray *prhs[], char *model_file_name)
|
143
|
-
{
|
144
|
-
int i, argc = 1;
|
145
|
-
char cmd[CMD_LEN];
|
146
|
-
char *argv[CMD_LEN/2];
|
147
|
-
void (*print_func)(const char *) = print_string_matlab; // default printing to matlab display
|
148
|
-
|
149
|
-
// default values
|
150
|
-
param.solver_type = L2R_L2LOSS_SVC_DUAL;
|
151
|
-
param.C = 1;
|
152
|
-
param.eps = INF; // see setting below
|
153
|
-
param.p = 0.1;
|
154
|
-
param.nr_weight = 0;
|
155
|
-
param.weight_label = NULL;
|
156
|
-
param.weight = NULL;
|
157
|
-
param.init_sol = NULL;
|
158
|
-
flag_cross_validation = 0;
|
159
|
-
col_format_flag = 0;
|
160
|
-
flag_C_specified = 0;
|
161
|
-
flag_solver_specified = 0;
|
162
|
-
flag_find_C = 0;
|
163
|
-
bias = -1;
|
164
|
-
|
165
|
-
|
166
|
-
if(nrhs <= 1)
|
167
|
-
return 1;
|
168
|
-
|
169
|
-
if(nrhs == 4)
|
170
|
-
{
|
171
|
-
mxGetString(prhs[3], cmd, mxGetN(prhs[3])+1);
|
172
|
-
if(strcmp(cmd, "col") == 0)
|
173
|
-
col_format_flag = 1;
|
174
|
-
}
|
175
|
-
|
176
|
-
// put options in argv[]
|
177
|
-
if(nrhs > 2)
|
178
|
-
{
|
179
|
-
mxGetString(prhs[2], cmd, mxGetN(prhs[2]) + 1);
|
180
|
-
if((argv[argc] = strtok(cmd, " ")) != NULL)
|
181
|
-
while((argv[++argc] = strtok(NULL, " ")) != NULL)
|
182
|
-
;
|
183
|
-
}
|
184
|
-
|
185
|
-
// parse options
|
186
|
-
for(i=1;i<argc;i++)
|
187
|
-
{
|
188
|
-
if(argv[i][0] != '-') break;
|
189
|
-
++i;
|
190
|
-
if(i>=argc && argv[i-1][1] != 'q' && argv[i-1][1] != 'C') // since options -q and -C have no parameter
|
191
|
-
return 1;
|
192
|
-
switch(argv[i-1][1])
|
193
|
-
{
|
194
|
-
case 's':
|
195
|
-
param.solver_type = atoi(argv[i]);
|
196
|
-
flag_solver_specified = 1;
|
197
|
-
break;
|
198
|
-
case 'c':
|
199
|
-
param.C = atof(argv[i]);
|
200
|
-
flag_C_specified = 1;
|
201
|
-
break;
|
202
|
-
case 'p':
|
203
|
-
param.p = atof(argv[i]);
|
204
|
-
break;
|
205
|
-
case 'e':
|
206
|
-
param.eps = atof(argv[i]);
|
207
|
-
break;
|
208
|
-
case 'B':
|
209
|
-
bias = atof(argv[i]);
|
210
|
-
break;
|
211
|
-
case 'v':
|
212
|
-
flag_cross_validation = 1;
|
213
|
-
nr_fold = atoi(argv[i]);
|
214
|
-
if(nr_fold < 2)
|
215
|
-
{
|
216
|
-
mexPrintf("n-fold cross validation: n must >= 2\n");
|
217
|
-
return 1;
|
218
|
-
}
|
219
|
-
break;
|
220
|
-
case 'w':
|
221
|
-
++param.nr_weight;
|
222
|
-
param.weight_label = (int *) realloc(param.weight_label,sizeof(int)*param.nr_weight);
|
223
|
-
param.weight = (double *) realloc(param.weight,sizeof(double)*param.nr_weight);
|
224
|
-
param.weight_label[param.nr_weight-1] = atoi(&argv[i-1][2]);
|
225
|
-
param.weight[param.nr_weight-1] = atof(argv[i]);
|
226
|
-
break;
|
227
|
-
case 'q':
|
228
|
-
print_func = &print_null;
|
229
|
-
i--;
|
230
|
-
break;
|
231
|
-
case 'C':
|
232
|
-
flag_find_C = 1;
|
233
|
-
i--;
|
234
|
-
break;
|
235
|
-
default:
|
236
|
-
mexPrintf("unknown option\n");
|
237
|
-
return 1;
|
238
|
-
}
|
239
|
-
}
|
240
|
-
|
241
|
-
set_print_string_function(print_func);
|
242
|
-
|
243
|
-
// default solver for parameter selection is L2R_L2LOSS_SVC
|
244
|
-
if(flag_find_C)
|
245
|
-
{
|
246
|
-
if(!flag_cross_validation)
|
247
|
-
nr_fold = 5;
|
248
|
-
if(!flag_solver_specified)
|
249
|
-
{
|
250
|
-
mexPrintf("Solver not specified. Using -s 2\n");
|
251
|
-
param.solver_type = L2R_L2LOSS_SVC;
|
252
|
-
}
|
253
|
-
else if(param.solver_type != L2R_LR && param.solver_type != L2R_L2LOSS_SVC)
|
254
|
-
{
|
255
|
-
mexPrintf("Warm-start parameter search only available for -s 0 and -s 2\n");
|
256
|
-
return 1;
|
257
|
-
}
|
258
|
-
}
|
259
|
-
|
260
|
-
if(param.eps == INF)
|
261
|
-
{
|
262
|
-
switch(param.solver_type)
|
263
|
-
{
|
264
|
-
case L2R_LR:
|
265
|
-
case L2R_L2LOSS_SVC:
|
266
|
-
param.eps = 0.01;
|
267
|
-
break;
|
268
|
-
case L2R_L2LOSS_SVR:
|
269
|
-
param.eps = 0.001;
|
270
|
-
break;
|
271
|
-
case L2R_L2LOSS_SVC_DUAL:
|
272
|
-
case L2R_L1LOSS_SVC_DUAL:
|
273
|
-
case MCSVM_CS:
|
274
|
-
case L2R_LR_DUAL:
|
275
|
-
param.eps = 0.1;
|
276
|
-
break;
|
277
|
-
case L1R_L2LOSS_SVC:
|
278
|
-
case L1R_LR:
|
279
|
-
param.eps = 0.01;
|
280
|
-
break;
|
281
|
-
case L2R_L1LOSS_SVR_DUAL:
|
282
|
-
case L2R_L2LOSS_SVR_DUAL:
|
283
|
-
param.eps = 0.1;
|
284
|
-
break;
|
285
|
-
}
|
286
|
-
}
|
287
|
-
return 0;
|
288
|
-
}
|
289
|
-
|
290
|
-
static void fake_answer(int nlhs, mxArray *plhs[])
|
291
|
-
{
|
292
|
-
int i;
|
293
|
-
for(i=0;i<nlhs;i++)
|
294
|
-
plhs[i] = mxCreateDoubleMatrix(0, 0, mxREAL);
|
295
|
-
}
|
296
|
-
|
297
|
-
int read_problem_sparse(const mxArray *label_vec, const mxArray *instance_mat)
|
298
|
-
{
|
299
|
-
mwIndex *ir, *jc, low, high, k;
|
300
|
-
// using size_t due to the output type of matlab functions
|
301
|
-
size_t i, j, l, elements, max_index, label_vector_row_num;
|
302
|
-
mwSize num_samples;
|
303
|
-
double *samples, *labels;
|
304
|
-
mxArray *instance_mat_col; // instance sparse matrix in column format
|
305
|
-
|
306
|
-
prob.x = NULL;
|
307
|
-
prob.y = NULL;
|
308
|
-
x_space = NULL;
|
309
|
-
|
310
|
-
if(col_format_flag)
|
311
|
-
instance_mat_col = (mxArray *)instance_mat;
|
312
|
-
else
|
313
|
-
{
|
314
|
-
// transpose instance matrix
|
315
|
-
mxArray *prhs[1], *plhs[1];
|
316
|
-
prhs[0] = mxDuplicateArray(instance_mat);
|
317
|
-
if(mexCallMATLAB(1, plhs, 1, prhs, "transpose"))
|
318
|
-
{
|
319
|
-
mexPrintf("Error: cannot transpose training instance matrix\n");
|
320
|
-
return -1;
|
321
|
-
}
|
322
|
-
instance_mat_col = plhs[0];
|
323
|
-
mxDestroyArray(prhs[0]);
|
324
|
-
}
|
325
|
-
|
326
|
-
// the number of instance
|
327
|
-
l = mxGetN(instance_mat_col);
|
328
|
-
label_vector_row_num = mxGetM(label_vec);
|
329
|
-
prob.l = (int) l;
|
330
|
-
|
331
|
-
if(label_vector_row_num!=l)
|
332
|
-
{
|
333
|
-
mexPrintf("Length of label vector does not match # of instances.\n");
|
334
|
-
return -1;
|
335
|
-
}
|
336
|
-
|
337
|
-
// each column is one instance
|
338
|
-
labels = mxGetPr(label_vec);
|
339
|
-
samples = mxGetPr(instance_mat_col);
|
340
|
-
ir = mxGetIr(instance_mat_col);
|
341
|
-
jc = mxGetJc(instance_mat_col);
|
342
|
-
|
343
|
-
num_samples = mxGetNzmax(instance_mat_col);
|
344
|
-
|
345
|
-
elements = num_samples + l*2;
|
346
|
-
max_index = mxGetM(instance_mat_col);
|
347
|
-
|
348
|
-
prob.y = Malloc(double, l);
|
349
|
-
prob.x = Malloc(struct feature_node*, l);
|
350
|
-
x_space = Malloc(struct feature_node, elements);
|
351
|
-
|
352
|
-
prob.bias=bias;
|
353
|
-
|
354
|
-
j = 0;
|
355
|
-
for(i=0;i<l;i++)
|
356
|
-
{
|
357
|
-
prob.x[i] = &x_space[j];
|
358
|
-
prob.y[i] = labels[i];
|
359
|
-
low = jc[i], high = jc[i+1];
|
360
|
-
for(k=low;k<high;k++)
|
361
|
-
{
|
362
|
-
x_space[j].index = (int) ir[k]+1;
|
363
|
-
x_space[j].value = samples[k];
|
364
|
-
j++;
|
365
|
-
}
|
366
|
-
if(prob.bias>=0)
|
367
|
-
{
|
368
|
-
x_space[j].index = (int) max_index+1;
|
369
|
-
x_space[j].value = prob.bias;
|
370
|
-
j++;
|
371
|
-
}
|
372
|
-
x_space[j++].index = -1;
|
373
|
-
}
|
374
|
-
|
375
|
-
if(prob.bias>=0)
|
376
|
-
prob.n = (int) max_index+1;
|
377
|
-
else
|
378
|
-
prob.n = (int) max_index;
|
379
|
-
|
380
|
-
return 0;
|
381
|
-
}
|
382
|
-
|
383
|
-
// Interface function of matlab
|
384
|
-
// now assume prhs[0]: label prhs[1]: features
|
385
|
-
void mexFunction( int nlhs, mxArray *plhs[],
|
386
|
-
int nrhs, const mxArray *prhs[] )
|
387
|
-
{
|
388
|
-
const char *error_msg;
|
389
|
-
// fix random seed to have same results for each run
|
390
|
-
// (for cross validation)
|
391
|
-
srand(1);
|
392
|
-
|
393
|
-
if(nlhs > 1)
|
394
|
-
{
|
395
|
-
exit_with_help();
|
396
|
-
fake_answer(nlhs, plhs);
|
397
|
-
return;
|
398
|
-
}
|
399
|
-
|
400
|
-
// Transform the input Matrix to libsvm format
|
401
|
-
if(nrhs > 1 && nrhs < 5)
|
402
|
-
{
|
403
|
-
int err=0;
|
404
|
-
|
405
|
-
if(!mxIsDouble(prhs[0]) || !mxIsDouble(prhs[1]))
|
406
|
-
{
|
407
|
-
mexPrintf("Error: label vector and instance matrix must be double\n");
|
408
|
-
fake_answer(nlhs, plhs);
|
409
|
-
return;
|
410
|
-
}
|
411
|
-
|
412
|
-
if(mxIsSparse(prhs[0]))
|
413
|
-
{
|
414
|
-
mexPrintf("Error: label vector should not be in sparse format");
|
415
|
-
fake_answer(nlhs, plhs);
|
416
|
-
return;
|
417
|
-
}
|
418
|
-
|
419
|
-
if(parse_command_line(nrhs, prhs, NULL))
|
420
|
-
{
|
421
|
-
exit_with_help();
|
422
|
-
destroy_param(¶m);
|
423
|
-
fake_answer(nlhs, plhs);
|
424
|
-
return;
|
425
|
-
}
|
426
|
-
|
427
|
-
if(mxIsSparse(prhs[1]))
|
428
|
-
err = read_problem_sparse(prhs[0], prhs[1]);
|
429
|
-
else
|
430
|
-
{
|
431
|
-
mexPrintf("Training_instance_matrix must be sparse; "
|
432
|
-
"use sparse(Training_instance_matrix) first\n");
|
433
|
-
destroy_param(¶m);
|
434
|
-
fake_answer(nlhs, plhs);
|
435
|
-
return;
|
436
|
-
}
|
437
|
-
|
438
|
-
// train's original code
|
439
|
-
error_msg = check_parameter(&prob, ¶m);
|
440
|
-
|
441
|
-
if(err || error_msg)
|
442
|
-
{
|
443
|
-
if (error_msg != NULL)
|
444
|
-
mexPrintf("Error: %s\n", error_msg);
|
445
|
-
destroy_param(¶m);
|
446
|
-
free(prob.y);
|
447
|
-
free(prob.x);
|
448
|
-
free(x_space);
|
449
|
-
fake_answer(nlhs, plhs);
|
450
|
-
return;
|
451
|
-
}
|
452
|
-
|
453
|
-
if (flag_find_C)
|
454
|
-
{
|
455
|
-
double best_C, best_rate, *ptr;
|
456
|
-
|
457
|
-
do_find_parameter_C(&best_C, &best_rate);
|
458
|
-
|
459
|
-
plhs[0] = mxCreateDoubleMatrix(2, 1, mxREAL);
|
460
|
-
ptr = mxGetPr(plhs[0]);
|
461
|
-
ptr[0] = best_C;
|
462
|
-
ptr[1] = best_rate;
|
463
|
-
}
|
464
|
-
else if(flag_cross_validation)
|
465
|
-
{
|
466
|
-
double *ptr;
|
467
|
-
plhs[0] = mxCreateDoubleMatrix(1, 1, mxREAL);
|
468
|
-
ptr = mxGetPr(plhs[0]);
|
469
|
-
ptr[0] = do_cross_validation();
|
470
|
-
}
|
471
|
-
else
|
472
|
-
{
|
473
|
-
const char *error_msg;
|
474
|
-
|
475
|
-
model_ = train(&prob, ¶m);
|
476
|
-
error_msg = model_to_matlab_structure(plhs, model_);
|
477
|
-
if(error_msg)
|
478
|
-
mexPrintf("Error: can't convert libsvm model to matrix structure: %s\n", error_msg);
|
479
|
-
free_and_destroy_model(&model_);
|
480
|
-
}
|
481
|
-
destroy_param(¶m);
|
482
|
-
free(prob.y);
|
483
|
-
free(prob.x);
|
484
|
-
free(x_space);
|
485
|
-
}
|
486
|
-
else
|
487
|
-
{
|
488
|
-
exit_with_help();
|
489
|
-
fake_answer(nlhs, plhs);
|
490
|
-
return;
|
491
|
-
}
|
492
|
-
}
|