-
-
Notifications
You must be signed in to change notification settings - Fork 1k
/
BinaryLabels.h
125 lines (110 loc) · 3.39 KB
/
BinaryLabels.h
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
/*
* This program 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 3 of the License, or
* (at your option) any later version.
*
* Written (W) 1999-2009 Soeren Sonnenburg
* Written (W) 1999-2008 Gunnar Raetsch
* Written (W) 2011-2012 Heiko Strathmann
* Copyright (C) 1999-2009 Fraunhofer Institute FIRST and Max-Planck-Society
*/
#ifndef _BINARY_LABELS__H__
#define _BINARY_LABELS__H__
#include <shogun/lib/config.h>
#include <shogun/lib/common.h>
#include <shogun/labels/LabelTypes.h>
#include <shogun/labels/DenseLabels.h>
namespace shogun
{
class CFile;
template <class T> class SGVector;
/** @brief Binary Labels for binary classification
*
* valid values for labels are +1/-1
*
* Scores may be converted into calibrated probabilities using
* scores_to_probabilities(), which implements the method described in
* Lin, H., Lin, C., and Weng, R. (2007).
* A note on Platt's probabilistic outputs for support vector machines.
* Should only be used in conjunction with SVM.
*/
class CBinaryLabels : public CDenseLabels
{
public:
/** default constructor */
CBinaryLabels();
/** constructor
*
* @param num_labels number of labels
*/
CBinaryLabels(int32_t num_labels);
#if !defined(SWIGJAVA) && !defined(SWIGCSHARP)
/** constructor
* sets labels with src elements
*
* @param src labels to set
*/
CBinaryLabels(SGVector<int32_t> src);
/** constructor
* sets labels with src elements (int64 version)
*
* @param src labels to set
*/
CBinaryLabels(SGVector<int64_t> src);
#endif
/** constructor
* sets values from src vector
* sets labels with sign of src elements with added threshold
*
* @param src labels to set
* @param threshold threshold
*/
CBinaryLabels(SGVector<float64_t> src, float64_t threshold = 0.0);
/** constructor
*
* @param loader File object via which to load data
*/
CBinaryLabels(CFile * loader);
/** Make sure the label is valid, otherwise raise SG_ERROR.
*
* possible with subset
*
* @param context optional message to convey the context
*/
virtual void ensure_valid(const char * context = NULL);
/** get label type
*
* @return label type binary
*/
virtual ELabelType get_label_type() const;
/** Converts all scores to calibrated probabilities by fitting a
* sigmoid function using the method described in
* Lin, H., Lin, C., and Weng, R. (2007).
* A note on Platt's probabilistic outputs for support vector machines.
*
* A sigmoid is fitted to the scores of the labels and then is used
* to compute porbabilities which are stored in the values vector. This
* is done via computing
* \f$pf=x*a+b\f$ for a given score \f$x\f$ and then computing
* \f$\frac{\exp(-f)}{1+}exp(-f)}\f$ if \f$f\geq 0\f$ and
* \f$\frac{1}{(1+\exp(f)}\f$ otherwise, where \f$a, bf\f$ are shape parameters
* of the sigmoid. These can be specified or learned automatically
*
* Should only be used in conjunction with SVM.
*
* @param a parameter a of sigmoid, if a=b=0, both are learned
* @param b parameter b of sigmoid, if a=b=0, both are learned
*/
void scores_to_probabilities(float64_t a = 0, float64_t b = 0);
/** @return object name */
virtual const char * get_name() const
{
return "BinaryLabels";
}
#ifndef SWIG // SWIG should skip this part
virtual CLabels* shallow_subset_copy();
#endif
};
}
#endif