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
|
// Copyright (C) 2002 Ronan Collobert (collober@iro.umontreal.ca)
//
//
// This file is part of Torch. Release II.
// [The Ultimate Machine Learning Library]
//
// Torch is free software; you can redistribute it and/or modify
// it under the terms of the GNU General Public License as published by
// the Free Software Foundation; either version 2 of the License, or
// (at your option) any later version.
//
// Torch is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
// GNU General Public License for more details.
//
// You should have received a copy of the GNU General Public License
// along with Torch; if not, write to the Free Software
// Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
#include "Boosting.h"
#include "BoostingMeasurer.h"
#include "random.h"
namespace Torch {
static void randw(int *selected_examples, real *ex_weights, int n_examples)
{
real *repartition = (real *)xalloc(sizeof(real)*(n_examples+1));
repartition[0] = 0;
for(int i = 0; i < n_examples; i++)
repartition[i+1] = repartition[i]+ex_weights[i];
for(int i = 0; i < n_examples; i++)
{
real z = uniform();
int gauche = 0;
int droite = n_examples;
while(gauche+1 != droite)
{
int centre = (gauche+droite)/2;
if(repartition[centre] < z)
gauche = centre;
else
droite = centre;
}
selected_examples[i] = gauche;
// printf("%g < %g < %g\n", repartition[gauche], z, repartition[gauche+1]);
}
free(repartition);
}
Boosting::Boosting(WeightedSumMachine* w_machine_, DataSet* data_, ClassFormat *class_format_) : Trainer(w_machine_, data_)
{
w_machine = w_machine_;
class_format = class_format_;
n_trainers = w_machine->n_trainers;
weights = w_machine->weights;
}
void Boosting::train(List* measurers)
{
int n_examples = data->n_examples;
int *selected_examples = (int *)xalloc(n_examples*sizeof(int));
real *ex_weights = (real *)xalloc(n_examples*sizeof(real));
for(int t = 0; t < n_examples; t++)
ex_weights[t] = 1./((real)n_examples);
BoostingMeasurer *measurer = new BoostingMeasurer(class_format, "/dev/null");
measurer->init();
measurer->setData(data);
measurer->setWeights(ex_weights);
message("Boosting: training...");
w_machine->n_trainers_trained = 0;
List *the_boost_mes = NULL;
addToList(&the_boost_mes, 1, measurer);
for(int i = 0; i < n_trainers; i++)
{
randw(selected_examples, ex_weights, n_examples);
data->pushSubset(selected_examples, n_examples);
w_machine->trainers[i]->machine->reset();
w_machine->trainers[i]->train(w_machine->trainers_measurers ? w_machine->trainers_measurers[i] : NULL);
data->popSubset();
// Find beta and all missclass
measurer->setInputs(w_machine->trainers[i]->machine->outputs);
w_machine->trainers[i]->test(the_boost_mes);
// Compute new weights
int *ptr_status = measurer->status;
real *ptr_ex_weights = ex_weights;
real beta = measurer->beta;
for(int t = 0; t < n_examples; t++)
{
if(*ptr_status++ > 0)
*ptr_ex_weights *= beta;
ptr_ex_weights++;
}
// Normalize
ptr_ex_weights = ex_weights;
real z = 0;
for(int t = 0; t < n_examples; t++)
z += *ptr_ex_weights++;
ptr_ex_weights = ex_weights;
for(int t = 0; t < n_examples; t++)
*ptr_ex_weights++ /= z;
// Warning: no precautions, no normalization
weights[i] = -log(beta);
w_machine->n_trainers_trained = i+1;
if(measurers)
test(measurers);
}
freeList(&the_boost_mes);
free(selected_examples);
free(ex_weights);
delete measurer;
}
Boosting::~Boosting()
{
}
}
|
